Allocation of Business Intelligence Costs · 2017-02-21 · the first day. Further, I would like to thank Prof. Dr. Ulrike Baumöl for taking on the co-supervision of this dissertation

Allocation of Business Intelligence Costs

DISSERTATION of the University of St. Gallen,

School of Management, Economics, Law, Social Sciences

and International Affairs to obtain the title of

Doctor of Philosophy in Management

submitted by

Johannes Michael Epple

rom

Germany

Approved on the application of Prof. Dr. Robert Winter

and

Prof. Dr. Ulrike Baumöl

Dissertation No. 4590

Difo-Druck GmbH, Bamberg 201

The University of St. Gallen, School of Management, Economics, Law, Social Sciences and International Affairs hereby consents to the printing of the present dissertation, with-out hereby expressing any opinion on the view herein expressed.

St. Gallen, October 24, 2016

The President:

Prof. Dr. Thomas Bieger

Acknowledgements i

Acknowledgements

The dissertation at hand is the achievement of a 3.5 year-long endeavor as an external PhD student at the Institute of Information Management at the University of St. Gallen (IWI-HSG). The successful completion of the dissertation would not have been possible without the support of certain people to whom I would like to express my gratitude at this point.

First of all, I would like to cordially thank my supervisor Prof. Dr. Robert Winter. His continuous support and guidance not only made this journey possible, but enriched it from a professional as well as from a personal point of view. In my role as an external PhD student, it was particularly helpful that he granted me the involvement in the re-search activities of the IWI-HSG, showed me opportunities along the way, made the expectations and responsibilities clear, and especially that he always kept his word from the first day. Further, I would like to thank Prof. Dr. Ulrike Baumöl for taking on the co-supervision of this dissertation. Her thoughtful and constructive comments always showed right directions and encouraged my work.

Besides my supervisor and my co-supervisor, I would like to thank special people whose willingness to discuss my ideas and selfless patience with me contributed a significant portion to this dissertation. My sincere thanks goes to Prof. Dr. Stephan Aier who pro-vided me with the opportunity to work on his projects and who always was a competent advisor with insightful comments. Dr. Stefan Bischoff deserves a place of honor in this paragraph for taking me by the hand on the PhD journey, for being available for feed-back at any time (even on very short notice), and especially for becoming a friend over the time. Further, a heartfelt thanks goes to Dr. Mohammad Kazem Haki who unfortu-nately only accompanied the last one and a half years of my PhD journey, but whose dedicated support meant a lot for this dissertation.

Especially, I also thank all my fellow PhD students and all other staff members from the chair of Prof. Dr. Robert Winter and the neighboring chairs. The exchange and experi-ences with you made the time in St. Gallen very joyful and memorable. To join the spirit among you and the mutual inspiration was a big treasure. Further, I thank all the co-authors with who I had the pleasure to directly work together.

Last but not least, I want to express my deepest gratitude to the most important people – my family, above all to my wife and our wonderful daughter. Taking up the challenge

ii Acknowledgements

of writing a PhD thesis meant going a long way paved with personal sacrifices and aus-terities – not only for myself, but especially for my personal environment. Therefore, I truly appreciate withstanding this time with me, the loving care of my closest, and their continuous backing that enabled me to do this.

St. Gallen, August 2016 Johannes Epple

Papers iii

Papers

Paper A: Epple, J., Bischoff, S., Winter, R., Aier, S. 2016. "Business Intelligence is no ‘Free Lunch’: What we Already Know About Cost Allocation for Busi-ness Intelligence Systems – and What we Should Find out", Working Pa-per, Institute of Information Management, University of St. Gallen, St. Gallen 2016, URL: https://www.alexandria.unisg.ch/publications/249525.

Paper B: Epple, J., Bischoff, S.; and Aier S. 2015. "Management Objectives and Design Principles for the Cost Allocation of Business Intelligence", in Proceedings of the Pacific Asia Conference on Information Systems (PACIS) 2015, Completed Research, Singapore, Singapore, Paper 187.

Paper C: Epple, J. 2016. "Contextual Factors Influencing the Purposeful Allocation of Business Intelligence Costs", in Proceedings of the European Confer-ence on Information Systems (ECIS) 2016, Completed Research, Istanbul, Turkey.

Paper D: Raber, D., Epple, J., Winter, R., Rothenberger, M. 2016. "Closing the Loop: Evaluating a Measurement Instrument for Maturity Model Design", in Proceedings of the Hawaii International Conference on System Sci-ences (HICSS) 2016, Completed Research, Koloa, Hawaii, USA, pp. 4444-4453.

Paper E: Epple, J., Fischer, E., Bischoff, S., Winter, R., and Aier, S. 2016. "Ignored, Accepted, or Used? Identifying the Phase of Acceptance of Business In-telligence Systems", in Multikonferenz Wirtschaftsinformatik (MKWI) 2016 – Band I, Completed Research, Nissen, V., Stelzer, D., Straßburger, S., and Fischer, D. (eds.), Universitätsverlag Ilmenau, Ilmenau, Germany, pp. 131-142.

Summary of Contents v

Summary of Contents

Acknowledgements ........................................................................................................ i

Papers ........................................................................................................................... iii

Summary of Contents .................................................................................................... v

Table of Contents ........................................................................................................ vii

List of Abbreviations ................................................................................................. xiii

List of Figures .............................................................................................................. xv

List of Tables ............................................................................................................. xvii

Abstract ...................................................................................................................... xix

Kurzfassung ............................................................................................................... xxi

Part A – Summary of the work .................................................................................... 1

1 Introduction .................................................................................................. 1

2 Conceptual Foundations ............................................................................ 10

3 Summary of Results ................................................................................... 17

4 Relationships Between Design Situations and Business Intelligence Cost Allocations ........................................................................................................... 25

5 Evaluation ................................................................................................... 36

6 Discussion .................................................................................................... 40

Part B – Papers of the Dissertation ............................................................................ 47

Paper A – Business Intelligence is no ‘Free Lunch’: What we Already Know About Cost Allocation for Business Intelligence Systems – and What we Should Find out ............................................................................................................ 47

Paper B – Management Objectives and Design Principles for the Cost Allocation of Business Intelligence ............................................................................. 61

Paper C – Contextual Factors Influencing the Purposeful Allocation of Business Intelligence Costs ......................................................................................... 83

Paper D – Closing the Loop: Evaluating a Measurement Instrument for Maturity Model Design ............................................................................................. 105

vi Summary of Contents

Paper E – Ignored, Accepted, or Used? Identifying the Phase of Acceptance of Business Intelligence Systems .............................................................................. 125

Bibliography............................................................................................................... cxli

Curriculum Vitae .................................................................................................... clxiii

Table of Contents vii

Table of Contents

Acknowledgements ........................................................................................................ i

Papers ........................................................................................................................... iii

Summary of Contents .................................................................................................... v

Table of Contents ........................................................................................................ vii

List of Abbreviations ................................................................................................. xiii

List of Figures .............................................................................................................. xv

List of Tables ............................................................................................................. xvii

Abstract ...................................................................................................................... xix

Kurzfassung ............................................................................................................... xxi

Part A – Summary of the work .................................................................................... 1

1 Introduction .................................................................................................. 1

1.1 Motivation ...................................................................................................... 1

1.2 Problem Description ....................................................................................... 2

1.3 Research Objective and Research Questions ................................................. 4

1.4 Research Approach ......................................................................................... 6

1.5 Structure of the Dissertation ........................................................................... 8

2 Conceptual Foundations ............................................................................ 10

2.1 Business Intelligence .................................................................................... 10

2.2 Cost Allocations ........................................................................................... 12

2.3 Related Work ................................................................................................ 14

3 Summary of Results ................................................................................... 17

3.1 Overview ...................................................................................................... 17

3.2 Papers of the Dissertation ............................................................................. 19

3.2.1 Paper A: Business Intelligence is no ‘Free Lunch’: What we Already Know About Cost Allocation for Business Intelligence Systems – and What we Should Find out ................................................. 19

viii Table of Contents

3.2.2 Paper B: Management Objectives and Design Principles for the Cost Allocation of Business Intelligence ................................................. 20

3.2.3 Paper C: Contextual Factors Influencing the Purposeful Allocation of Business Intelligence Costs ................................................................. 21

3.2.4 Paper D: Closing the Loop: Evaluating a Measurement Instrument for Maturity Model Design ...................................................................... 22

3.2.5 Paper E: Ignored, Accepted, or Used? Identifying the Phase of Acceptance of Business Intelligence Systems ......................................... 23

4 Relationships Between Design Situations and Business Intelligence Cost Allocations ........................................................................................................... 25

4.1 Business Intelligence Cost Allocations in Archetypical Design Situations ...................................................................................................... 25

4.2 Propositions for the Configuration of Business Intelligence Cost Allocations ................................................................................................... 30

4.3 Procedure for Business Intelligence Cost Allocations ................................. 33

5 Evaluation ................................................................................................... 36

5.1 Evaluation of Solution Components ............................................................ 36

5.2 Overall Evaluation ....................................................................................... 37

6 Discussion .................................................................................................... 40

6.1 Summary and Limitations ............................................................................ 40

6.2 Implications for Practice and Research ........................................................ 43

6.3 Fields for Future Research ........................................................................... 44

Part B – Papers of the Dissertation............................................................................ 47

Paper A – Business Intelligence is no ‘Free Lunch’: What we Already Know About Cost Allocation for Business Intelligence Systems – and What we Should Find out ........................................................................................................... 47

A.1 Introduction .................................................................................................. 48

A.2 Conceptual Foundations ............................................................................... 49

A.3 Types of literature reviews ........................................................................... 50

A.4 Literature Review ......................................................................................... 51

Table of Contents ix

A.4.1 First Iteration ............................................................................................ 52

A.4.2 Second Iteration ........................................................................................ 54

A.4.3 Third Iteration ........................................................................................... 55

A.5 Literature Synthesis and Future Research .................................................... 56

A.5.1 Literature synthesis ................................................................................... 56

A.5.2 Fields for future research .......................................................................... 57

A.6 Conclusion .................................................................................................... 59

Paper B – Management Objectives and Design Principles for the Cost Allocation of Business Intelligence ............................................................................. 61

B.1 Introduction .................................................................................................. 62

B.2 Related Work ................................................................................................ 63

B.3 Research Method .......................................................................................... 65

B.3.1 Focus Group ............................................................................................. 66

B.3.2 Design Principles ...................................................................................... 67

B.4 Exploratory Focus Group ............................................................................. 67

B.4.1 Focus Group Setup and Procedure ........................................................... 67

B.4.2 Management Objectives of BI Cost Allocations ...................................... 70

B.4.3 Design Principles for BI Cost Allocations ............................................... 72

B.5 Confirmatory Focus Group .......................................................................... 75

B.5.1 Focus Group Setup and Procedure ........................................................... 75

B.5.2 Evaluation of Management Objectives .................................................... 77

B.5.3 Evaluation of Design Principles ............................................................... 78

B.6 Discussion and Conclusion .......................................................................... 79

Paper C – Contextual Factors Influencing the Purposeful Allocation of Business Intelligence Costs ......................................................................................... 83

C.1 Introduction .................................................................................................. 84

C.2 Conceptual Foundation ................................................................................. 86

C.2.1 Cost Allocations ....................................................................................... 86

x Table of Contents

C.2.2 Business Intelligence ................................................................................ 86

C.2.3 Situational Context ................................................................................... 87

C.2.4 Research Gap ........................................................................................... 88

C.3 Research Method .......................................................................................... 89

C.3.1 Research Process ...................................................................................... 89

C.3.2 Literature Review ..................................................................................... 90

C.3.3 Focus Group ............................................................................................. 91

C.4 Results of Literature Review ........................................................................ 93

C.4.1 Management Objectives ........................................................................... 93

C.4.2 Contextual Factors ................................................................................... 95

C.5 Results of Focus Group ................................................................................ 98

C.6 Synthesis of Results ..................................................................................... 99

C.7 Discussion and Conclusion ........................................................................ 103

Paper D – Closing the Loop: Evaluating a Measurement Instrument for Maturity Model Design ............................................................................................. 105

D.1 Introduction ................................................................................................ 106

D.2 Prior Work .................................................................................................. 107

D.2.1 Existing Maturity Models ...................................................................... 107

D.2.2 Development of the BI maturity model and the measurement instrument ............................................................................................... 109

D.2.3 Refinement of Measurement Instrument ................................................ 110

D.3 Evaluation Approach .................................................................................. 110

D.3.1 Selection of Maturity Clusters for Evaluation ....................................... 110

D.3.2 Data Collection and Analysis ................................................................. 112

D.4 Evaluation................................................................................................... 114

D.4.1 Empirical Setting .................................................................................... 114

D.4.2 Results .................................................................................................... 115

D.4.2.1 Company A ........................................................................................ 115

Table of Contents xi

D.4.2.2 Company B ......................................................................................... 117

D.4.2.3 Company C ......................................................................................... 119

D.5 Discussion .................................................................................................. 121

D.6 Limitations and Future Research ................................................................ 122

Paper E – Ignored, Accepted, or Used? Identifying the Phase of Acceptance of Business Intelligence Systems ............................................................................... 125

E.1 Introduction ................................................................................................ 126

E.2 Conceptual Foundations ............................................................................. 127

E.3 Antecedents for the Phases of Acceptance ................................................. 128

E.3.1 Internal Antecedents ............................................................................... 129

E.3.2 External Antecedents .............................................................................. 132

E.3.3 Summary of Antecedents ....................................................................... 134

E.4 Case Study .................................................................................................. 134

E.4.1 Case Description and Empirical Setting ................................................. 134

E.4.2 Data Analysis ......................................................................................... 136

E.5 Discussion and Implications on BI Management ....................................... 138

E.6 Limitations and Conclusion........................................................................ 139

Bibliography ............................................................................................................... cxli

Curriculum Vitae .................................................................................................... clxiii

List of Abbreviations xiii

List of Abbreviations

ABC Affect, Behavior, Cognition (in paper E)

ABC Activity Based Costing (in part A and the other papers)

AIS Association for Information Systems

BI Business Intelligence

BICC Business Intelligence Competence Center

BIMM Business Intelligence Maturity Model

BIS Business Intelligence System(s)

CA(s) Cost Allocation(s)

CEO Chief Executive Officer

CFO Chief Financial Officer

CFG Confirmatory Focus Group

CIO Chief Information Officer

COBIT Control Objectives for Information and Related Technologies

DSR Design Science Research

DWH Data Warehouse

ECIS European Conference on Information Systems

EFG Exploratory Focus Group

ERP enterprise resource planning

FG Focus Group

HICSS Hawaii International Conference on System Sciences

IMap Intrinsic Motivation Toward Accomplishment

IMkw Intrinsic Motivation to Know

IMst Intrinsic Motivation to Experience Stimulation

IP Interview Partner

IS Information System(s)

xiv List of Abbreviations

IT Information Technology

MI Measurement Instrument

MIS Management Information System(s)

MKWI Multikonferenz Wirtschaftsinformatik

MM(s) Maturity Model(s)

OLAP Online Analytical Processing

PACIS Pacific Asia Conference on Information Systems

PEU Perceived Ease of Use

PU Perceived Usefulness

RIM Rich Intrinsic Motivation

RQ Research Question

SME Situational Method Engineering

TR TR Telecom

VHB Verband der Hochschullehrer für Betriebswirtschaft

List of Figures xv

List of Figures

Figure 1: Different Approaches to Define a Business Intelligence Cost Allocation ...... 3

Figure 2: Interrelation of Research Questions ................................................................. 4

Figure 3: Basic Cost Allocation Mechanisms ............................................................... 13

Figure 4: Overall Conceptual Design ............................................................................ 17

Figure 5: Procedure for the Configuration of Business Intelligence Cost Allocations . 35

Figure 6. Radar Charts Illustrating the Four Different Maturity Clusters ................... 112

Figure 7. Radar Charts of Results from Alternative Maturity Assessments ............... 121

Figure 8: Phases of Acceptance and Stage Model of IS Implementation ................... 128

List of Tables xvii

List of Tables

Table 1: Research Questions ........................................................................................... 4

Table 2: How the Research Questions are Addressed ................................................... 18

Table 3: Design Situation of the Running Blind BI Company...................................... 27

Table 4: Design Situation of the Stagnant BI Company ............................................... 28

Table 5: Design Situation of the Adaptive BI Company ............................................... 29

Table 6: Dominating Contextual Factors ...................................................................... 33

Table 7: Contributions Related to Research Questions and Papers .............................. 41

Table 8: Bibliographical Information for Paper A ........................................................ 47

Table 9. Results of First Search Iteration ...................................................................... 53

Table 10. Analysis of First Search Iteration .................................................................. 53

Table 11. Results of Second Search Iteration ................................................................ 54

Table 12. Analysis of Second Search Iteration ............................................................. 54

Table 13: Bibliographical Information for Paper B ...................................................... 61

Table 14: Participants of EFG ....................................................................................... 69

Table 15: Company Profiles of Participants of EFG ..................................................... 70

Table 16: Principles of Form for a BI Cost Allocation Method .................................... 73

Table 17: Principles of Function for a BI Cost Allocation Method .............................. 74

Table 18: Participants and Company Profiles of CFG .................................................. 77

Table 19: Evaluation of Management Objectives ......................................................... 78

Table 20: Design Principles for a BI Cost Allocation Method ..................................... 79

Table 21: Bibliographical Information for Paper C ...................................................... 83

Table 22: Basic Cost Allocation Mechanisms .............................................................. 86

Table 23: Focus Group Participants .............................................................................. 92

Table 24: Overview of Management Objectives ........................................................... 94

Table 25: Contextual Factors Related to Technology ................................................... 95

Table 26: Contextual Factors Related to Organization ................................................. 96

xviii List of Tables

Table 27: Contextual Factors Related to Environment ................................................. 97

Table 28: Morphological Box Comprising Relevant Management Objectives and Contextual Factors ....................................................................................................... 100

Table 29: Bibliographical Information for Paper D .................................................... 105

Table 30. Overview of Existing BI Maturity Models ................................................. 108

Table 31: Overview of Companies .............................................................................. 114

Table 32: Overview of Interviewees ........................................................................... 115

Table 33: Comparison of Maturity Assessments for Company A .............................. 117

Table 34: Comparison of Maturity Assessments for Company B .............................. 119

Table 35: Comparison of Maturity Assessments for Company C .............................. 120

Table 36: Bibliographical Information for Paper E .................................................... 125

Table 37: Overview of Beliefs Across all Acceptance Phases ................................... 132

Table 38: Overview of Interviewees ........................................................................... 135

Table 39: Results of Data Analysis ............................................................................. 137

Abstract xix

Abstract

Along with the pervasion of business intelligence (BI) in organizations, cost-intensive investments for establishing and maintaining BI are made. Today, BI has become widely established from a technological perspective, but remains challenging from an organi-zational view. The high cost block, which is caused by BI, often remains as opaque and unmanageable overhead costs. Therefore, appropriate cost management mechanisms need to be in place to purposefully manage BI costs. Cost allocations (CAs) for BI sys-tems are intended to support the business-information technology (IT) alignment, en-hance transparency, create cost awareness, and support the management of BI system resources. Although CA for BI systems is highly relevant for practice, the field is widely unexplored in the current scientific literature. Prior work on CAs of information systems (IS) and BI costs show that a “one size fits all” approach for allocating BI costs is not presumed to be purposeful; rather, CAs need to be configured with respect to the design situation, which is characterized by management objectives and contextual factors.

This cumulative dissertation follows the design science research paradigm and provides a set of instruments that simplifies and guides the design of optimal CA for BI costs while incorporating the specific design situation. As such, first, design principles are set up that form the basis for BI CAs independent from design situations. Second, the man-agement objectives and different contextual factors characterizing the design situations are specified. Further, systematic instruments to identify the states of the contextual fac-tors BI acceptance and BI maturity in organizations are developed. Third, the interrela-tions between relevant design situations and the configuration of BI CAs are demon-strated and propositions for the configuration of BI CAs are derived, which add guidance on how BI CAs should be configured in certain design situations.

Therefore, the results of this dissertation have manifold implications. For practice, it proposes valuable insights that shall be considered in the design process of a BI CA. These insights lead to a more efficient process of finding the optimal CA for BI costs. For research, the dissertation at hand contributes to the existing descriptive and prescrip-tive knowledge base by bridging the identified research gaps regarding specific research phenomena and by delivering a compelling solution approach to a current design prob-lem. In addition, it stimulates further research on context-specific BI cost management.

Keywords: Business Intelligence, BI Management, IS Management, BI Governance, Cost Accounting, Cost Management, Business-IT Alignment, Contextual Factors.

Kurzfassung xxi

Kurzfassung

Mit dem Einsatz von Business Intelligence (BI) in Unternehmen gehen hohe Investitio-nen für die Entwicklung und den Betrieb von BI einher. Die BI Technologie ist heute weitgehend etabliert, allerdings bestehen Herausforderungen in Bezug auf die Organi-sation von BI. Der hohe Kostenblock, den BI verursacht, verbleibt oft als intransparente und nicht steuerbare Gemeinkosten. Deswegen müssen geeignete Mechanismen einge-setzt werden, um eine Steuerung der BI Kosten zu ermöglichen. Kostenverrechnungen für BI unterstützen das Business-IT Alignment, erhöhen die Kostentransparenz, schaf-fen Kostenbewusstsein und unterstützen die Steuerung des Ressourceneinsatzes. Bishe-rige Arbeiten zeigen, dass ein “one size fits all” Ansatz nicht zielführend ist, sondern BI Kostenverrechnungen unter Beachtung der jeweiligen Designsituation zu konfigurieren sind.

Die vorliegende kumulative Dissertation folgt dem Paradigma der gestaltungsorientier-ten Forschung und stellt ein Instrumentarium bereit, das die Gestaltung optimaler Kos-tenverrechnungen für BI Kosten anleitet und vereinfacht, wobei die jeweiligen Design-situation berücksichtigt werden. Dazu werden in einem ersten Schritt Designprinzipien aufgestellt, die unabhängig von der Designsituation die Basis für BI Kostenverrechnun-gen darstellen. Im zweiten Schritt werden die Managementziele und Kontextfaktoren, die die Designsituationen charakterisieren, spezifiziert. Außerdem werden zwei syste-matische Instrumente zur Identifizierung der Kontextfaktoren BI Akzeptanz und BI Rei-fegrad entwickelt. Im dritten Schritt werden Zusammenhänge zwischen relevanten De-signsituationen und der Konfiguration von BI Kostenverrechnungen dargelegt und Ge-staltungsempfehlungen für BI Kostenverrechnungen abgeleitet, wie die Konfiguration von BI Kostenverrechnungen in bestimmten Designsituationen zu gestalten ist.

In dieser Dissertation werden für den Einsatz in der Praxis wertvolle Erkenntnisse ge-wonnen, die in der Gestaltung einer BI Kostenverrechnung berücksichtigt werden soll-ten. Diese Erkenntnisse führen zu Effizienzsteigerungen im Findungsprozess einer op-timalen BI Kostenverrechnung. Aus wissenschaftlicher Sicht leistet die Dissertation durch Schließen der identifizierten Forschungslücke sowie durch stringente Lösungsan-sätze ihren Beitrag zur deskriptiven und präskriptiven Wissensbasis. Zudem werden weitere Forschungsfelder für kontextabhängiges BI Kostenmanagement aufgezeigt.

Stichwörter: Business Intelligence, BI Management, IS Management, BI Governance, Kostenrechnung, Kostenmanagement, Business-IT Alignment, Kontextfaktoren.

Part A: Introduction 1

Part A – Summary of the work

1 Introduction

1.1 Motivation Together with the pervasion of transactional information systems (IS) in organizations, business intelligence (BI) technologies have developed that can transform vast amounts of existing data into decision support information. Today, BI is established as an um-brella term for “technologies, applications and processes for gathering, storing, access-ing and analyzing data to help its users make better decisions” (Wixom and Watson 2010, p. 14).

While expenses for information technology (IT) only grew 0.4% in 2013 (Gartner Inc. 2014a), BI investments rose by 8% in 2013 (Gartner Inc. 2014b) despite cost pressure and declining budgets in administrative departments. This development highlights the growing importance of BI in practice. One reason for increasing expenses for BI is the exponential growth of data, which inevitably leads to increasing costs for storage and analysis. However, Brynjolfsson and Hitt (1998, p. 55) already showed that higher IT investments do not necessarily lead to increased productivity. Furthermore, in times of ever increasing cost pressure on IT and, therefore on BI departments, the efficiency of BI departments needs to be continuously demonstrated, performance evaluated, and po-tentials for cost optimization revealed. As a consequence, organizations need to trans-parently present and justify the high expenses for BI. A cost allocation (CA) for BI is supposed to enhance transparency for this growing portion of overhead costs and enable organizations to allocate costs according to their source.

For IS research and, in particular, for IS governance research, the topic of CAs that allocate IS costs from an IS-providing unit to an IS-consuming unit is a core activity of IS cost management (Van Grembergen and De Haes 2009, pp. 124-125). CAs are sup-posed to bring several benefits to an organization and fulfil various purposes, e.g., to enhance cost transparency, to inform investment decisions, to provide a basis for product calculation purposes, to create cost consciousness, or to reveal inefficiencies in the use of resources (Kaplan and Cooper 1998, p. 1; Klesse 2007, pp. 37-38; Shim and Siegel 2000, p. 75). Therefore, a BI CA is an appropriate instrument to achieve the desired effects for managing increasing BI costs.

2 Part A: Introduction

Due to the special characteristics of BI, the costs involved appear as a special subject of CA and differ from other internally provided services within an organization (cf., Sec-tion 2.1). A review of prior work (cf., Section 2.3) shows that several publications call for future research on BI cost management in particular. Moreover, the maturity of re-search and practical application of IS CA is considered to be still low (Stefanov et al. 2012, p. 1), and it can be observed that BI CAs need to be configured in respect of the different design situations prevailing in organizations. Put differently, there is no “one size fits all” approach for the configuration of BI CAs. To conclude, BI CA is a highly relevant topic for theory and practice, but is not yet sufficiently explored regarding the different design situations.

To address the above mentioned issues, this dissertation follows the design science re-search (DSR) paradigm (Hevner et al. 2004) and contributes theoretically founded guid-ance that can be applied in practice to find purposeful BI CAs in certain design situa-tions. The identified research gaps, the particularities of BI in regard to CAs, as well as the practical relevance comprising high prospective benefits for organizations by imple-menting purposeful BI CAs, motivate this dissertation.

1.2 Problem Description To contribute results that are scientifically rigorous as well as practically relevant (Hevner and Chatterjee 2010, p. 12; Stokes 1997), the playing field needs to be delimited and the research problem denominated. Therefore, in the following the research problem is described from both a theoretical and practical perspective.

The state of the art in scientific literature regarding BI CAs (cf., Section 2.3) reveals that on one hand there is a high need for proper cost management, cost transparency, and cost-value considerations in the domain of BI. On the other hand, no research-based publications deal with the topic of BI CAs in a comprehensive and compelling way. Several sources provide examples of specific configurations of IS or BI CAs in certain situational contexts (e.g., Grytz 2014; Rosenkranz and Holten 2007; Watson et al. 2004), but no work comprehensively deals with the purposeful configuration of IS or BI CAs in different design situations, and the instantiation to BI is missing (cf., paper A, see pp. 47-60). Existing CA approaches are not easily transferrable and applicable to BI, which is characterized by its special nature in regard to CAs (cf., pp. 10-12).

The practical relevance of the topic is derived from existing practitioner-oriented publi-cations (see Section 2.3) as well as from contact with industry partners (cf., exploratory focus group in paper B, see pp. 67-75). Several archetypes of CA mechanisms exist,


which need to be further adapted to BI. The adaptation of a basic CA mechanism to BI is regarded as the configuration. Since the configuration of BI CAs is a complex task and theoretically informed guidance for practice is lacking, organizations tend to be in-efficient in their configuration processes to find the optimal BI CA. Figure 1 visualizes different approaches to find the optimal BI CA and the missing approach, which can be considered as the research problem from a practical point of view. Without appropriate guidance, organizations either find the optimal BI CA by trial and error or through a lengthy endeavor. In the trial and error approach, the definition of the BI CA is effi-ciently conducted, but leads to an inferior solution. Therefore, the definition is continu-ously reconfigured as often as necessary until an effective BI CA is found. On the con-trary, in the lengthy endeavor approach, a long lasting inefficient search process for finding the optimal BI CA is conducted due to the missing scientifically grounded guid-ance. In the existing approaches, organizations tend to take inferior decisions on the configuration of BI CAs, e.g., due to missing guidance, missing experience, or gut feel-ings. Therefore, the research problem from a practical standpoint is the missing guidance for practice to efficiently determine an optimal BI CA.

Figure 1: Different Approaches to Define a Business Intelligence Cost Allocation


1.3 Research Objective and Research Questions Pursuant to the above described research problem, in this section the research objective as well as the research questions (RQs) are derived. As described in the motivation (see Section 1.1), it is accepted that there is no “one size fits all” approach to BI CAs. There-fore, this dissertation does not strive for a “one size fits all” solution, but for

… a set of instruments that simplifies, systematizes, and guides the quest for purpose-ful BI CAs.

To close the existing research gaps and to support practitioners with an efficient search process for BI CAs, the solution approach needs to incorporate the specific design situ-ation. This leads to the RQs of this dissertation, which are summarized in Table 1. The RQs are built upon each other in order to contribute to the overall research objective. Figure 2 visualizes the interrelation between the RQs. Subsequently, the RQs are further elucidated in detail.

Table 1: Research Questions

1. Design principles for business intelligence cost allocations RQ 1 What are the design principles for business intelligence cost allocations?

2. Design situations for business intelligence cost allocations RQ 2 What characterizes relevant design situations for business intelligence cost

allocations? 3. Configuration of business intelligence cost allocations

RQ 3 How should business intelligence cost allocations be configured in certain design situations?

Figure 2: Interrelation of Research Questions


RQ 1: What are the design principles for business intelligence cost allocations?

RQ 1 specifically targets the research gap between the theoretical and practical rele-vance of BI CAs and missing solution approaches and artifacts. Whereas publications in scientific outlets highlight the need for future research, practitioner-oriented publica-tions do not deliver compelling answers from practice, but state that the maturity of the topic is still low (cf., pp. 47-60). Further, BI experts from industry partners (cf., focus group studies in paper B, see pp. 67-75) face challenges related to the topic and call for guidance. RQ 1 results from this dilemma and aims at providing research and practice with design principles (Gregor et al. 2013, p. 4) for BI CAs, though independent from the specific situational context. In general, design principles give advice for effectively designing and implementing solutions in specific contexts. To pursue this goal, certain design principles, which are universally valid for BI CAs, shall constitute the foundation on which BI CAs can be built in different situational contexts. Due to missing prior work in the field of BI CAs, the developed design principles represent the common ground of this dissertation.

RQ 2: What characterizes relevant design situations for business intelligence cost allo-cations?

RQ 2 refers to the assumption that there is no “one size fits all” solution for the design of BI CAs. Every BI CA is designed for a specific purpose and in a given situational context. Therefore, RQ 2 aims at identifying the relevant dimensions that characterize design situations for BI CAs.

A BI CA is not an end in itself, but is employed for certain purposes, i.e., to achieve or fulfil management objectives. Hence, the management objectives for BI CAs are the subject of this RQ and as they represent one dimension of the design situation. The management objectives are one important constituent of design situations for BI CAs since they are the motivation and justification for designing BI CAs.

Further, the design situations of BI CAs are supposed to be characterized by various contextual factors, e.g., the level of BI maturity. Thus, contextual factors are – beside the management objectives – the second major dimension constituting the design situa-tion of BI CAs. Therefore, the further purpose of this RQ is to identify and assess dif-ferent contextual factors and their characteristics that need to be incorporated in BI CAs. To assess the contextual factors, instruments that determine the characteristics of the contextual factors need to be available, e.g., an instrument to determine the BI maturity level within an organization. Therefore, the purpose of this RQ is not only to identify


management objective and contextual factors, but also to provide instruments to deter-mine the characteristics of relevant contextual factors.

RQ 3: How should business intelligence cost allocations be configured in certain design situations?

RQ 3 aims at the actual design of BI CAs under consideration of the different design situations defined in RQ 2. The answers to RQ 3 are intended to uncover relations be-tween the design situations and the configurations of BI CAs.

Therefore, for answering RQ 3, certain design situations resulting from the answers to RQ 2 are supposed to be identified. By further analyzing the identified design situations, impacts of the design situations or its constitutive dimensions, respectively, and the con-figurations of BI CAs are revealed. Further, the obtained insights shall be practically relevant, but concurrently generalizable to a certain extent. The result of RQ 3 contrib-utes to closing the identified research gap regarding BI CAs by providing insight on the relations between design situations and configurations of BI CAs as well as providing necessary steps for solving existing real-world BI CA design problems. The derived results shall be rigorously researched and support practitioners (especially BI managers and cost accountants) in the purposeful design of BI CAs.

To contribute a useful solution to solve the existing real-world design problem this dis-sertation follows the DSR paradigm (Hevner and Chatterjee 2010; Peffers et al. 2007). Therefore, the following section introduces the overall research approach.

1.4 Research Approach In this section, the DSR paradigm according to Peffers et al. (2007) is introduced as the overall research method. Due to the cumulative nature of the dissertation at hand, the single papers do not necessarily strictly follow the DSR approach, but individually ap-propriate research methods are applied. The individual research methods of the single publications are presented in the respective papers (cf., pp. 47-140).

In contrast to behavioral research, which aims at exploratory and descriptive learning (Winter and Baskerville 2010, p. 269), DSR aims at “prescriptive learning, i.e., the de-sign and evaluation of innovative, useful, generic problem solutions to important and relevant design problems in organizations” (Winter and Baskerville 2010, p. 269). Therefore, DSR intends to solve existing real-world design problems by providing use-ful artifacts (Hevner et al. 2004, p. 82; March and Smith 1995, p. 160; Simon 1996, p. 114) instead of understanding existing phenomena (Hevner et al. 2004, p. 82). Different


artifact types need to be distinguished in DSR, such as constructs, models, methods, and instantiations (March and Smith 1995, p. 253; Winter 2008, p. 471) . Gregor and Hevner (2013, p. 341) add further artifact types, namely, design theories, design principles, and technological rules. Due to the cumulative nature of this dissertation, the result is not “one” artifact, but a suite of useful artifacts – stemming from the different papers – that support the overall research goal. Therefore, this dissertation is cut down into several sub-topics, whose results are published stepwise and summarized in part A of this dis-sertation.

Peffers et al. (2007, p. 54) propose a methodology for DSR comprising six activities: identify problem and motivate; define objectives of a solution; design and development; demonstration; evaluation; and communication. These six activities should not be un-derstood as a pure linear process, but in an iterative manner especially from the activities evaluation and communication back to the activities that define objectives of a solution and design and development. Due to the cumulative nature of this dissertation, the single phases of the DSR process cannot be clearly delineated. The single papers are, to some extent, single DSR projects in themselves that apply the introduced research paradigm to a certain extent by working out and evaluating results; however, not every single paper runs through the entire process and all six phases. The communication phase of the build-evaluate iterations runs concurrently to the remaining activities by publishing the papers in IS journals and presenting at IS conferences.

In contrast to Peffers et al. (2007, p. 56), who call for ex post evaluations of the designed artifact, other authors (e.g., Abraham et al. 2014; Sonnenberg and vom Brocke 2012; Venable et al. 2012) claim continuous evaluation activities during the design process. Sonnenberg and vom Brocke (2012) stress the iterative manner of the DSR process, and propose a design process consisting of several build-evaluate circles with four interme-diate evaluations – two ex ante and two ex post evaluations. During the course of this dissertation, design and evaluation activities have been alternated according to Sonnen-berg and vom Brocke (2012), although the proposed research project does not stick to exactly four intermediate evaluations. Due to the high importance of evaluations in DSR, the conducted evaluation activities are described in Section 5.

Besides the presented research method applied in this dissertation, the theoretical foun-dations for different design situations, in which BI CAs are implemented, is introduced subsequently. Therefore, the concepts of situational method engineering (SME) (Kumar and Welke 1992; Ralyté et al. 2007) and contingency theory (Weill and Olson 1989) are briefly introduced for this dissertation. The adoption of contingency theory and SME to


IS research has taken place in a wide range of publications (e.g., Brinkkemper 1996; Bucher et al. 2007; Reid and Smith 2000; Schonberger 1980). The credo of contingency theory is that “there is no one best way to organize” (Galbraith 1973, p. 2), but there are several effective ways, depending on the situational context, in which the organization is embedded. Varying organizational contexts can be defined by contingency variables (Weill and Olson 1989, pp. 60-65), which are also known as contextual factors. To take the specific design situations of BI CAs into account, and not to design a “one size fits all” solution for BI CAs, the conception of SME is partially applied. This dissertation follows the distinction made by Bucher et al. (2007, pp. 33-41) characterizing design situations and artifact fragments as central components of SME. On one hand, the design of BI CAs is highly dependent on the design situation, in which the allocation should be designed. On the other hand, in every BI CA, various components (the artifact frag-ments) have to be configured according to the design situation. Therefore, RQ 2 aims at identifying the contextual factors and the management objectives constituting the design situations of BI CAs. The fragments of BI CAs that need to be configured in accordance with the specific design situations are provided through the existing general CA mech-anisms presented in Section 2.2. However, the results are not derived in a pure SME project, but rather SME enriches the research process with valuable conceptions and insights for target-oriented efforts toward the desired results.

1.5 Structure of the Dissertation The dissertation at hand consists of two parts. In part A (the synopsis paper at hand), the overall research results are summarized. Part B contains the five single research papers, which contribute individual pieces of research to the overall research results.

Part A is organized as follows: section one presents the motivation behind this research, the description of the research problem, the research objective with its derived research questions, the overall research approach, and this section on the structure of the work at hand. Section two lays conceptual foundations by introducing the understanding of BI in this dissertation from an IS perspective, relevant CA mechanisms from an accounting perspective, and a brief overview of related work. In section three the five papers are related to the overall research and the papers are summarily delineated. In section four, relations between different design situations and the different configurations of BI CAs are depicted and propositions for the configuration of BI CAs are presented. Section five outlines the single evaluation steps conducted in the course of the research endeavors of the five papers as well as the rationale for the overall evaluation. Section six critically


discusses the results, presents its limitations, and illustrates potential directions for fu-ture research.

In part B, the corresponding full texts of the single papers are included. For consistency reasons, the papers are presented in a uniform layout. Therefore, a consistent citation style is used throughout this dissertation. The tables and figures are continuously num-bered and their layouts in the text might be slightly deviating from the published ver-sions of the papers. The work at hand, contains a comprehensive list of figures, list of tables, and list of abbreviations at the beginning of this document, and a comprehensive bibliography at the end comprising all references. Further, prior to each paper in part B, the corresponding bibliographical metadata, the abstract, and keywords (if existing) are introduced.

10 Part A: Conceptual Foundations

2 Conceptual Foundations

In this section, the concept of BI is outlined in Section 2.1, since it is the domain to which CA is applied in this dissertation. Afterwards, in Section 2.2, CAs commonly recognized in the field of cost accounting are introduced because BI CAs are conceptu-ally located at the interface between accounting and IS. In any case, IS is an interdisci-plinary field. Consequently, literature from other disciplines needs to be considered for theoretical advancements in IS (Rowe 2014, p. 247; Webster and Watson 2002, p. xvi).

2.1 Business Intelligence In 1958, Hans Peter Luhn first described BI and BI systems as means to “accommodate all information problems of an organization” (Luhn 1958, p. 314). Ever since, various concepts with slightly altering scopes have been developed for IS enabled decision sup-port. The most commonly known among them are management information systems (Gallagher 1961), decision support systems (French and Turoff 2007) , and executive information systems (Rockart and Treacy 1980, p. 3).

While in the early development stages the IS for decision support had rather narrow perspectives, today the concept of BI is usually understood in a broader sense, since it encompasses all components of an integrated decision support infrastructure (Baars and Kemper 2008, p. 132). This comprises the technological as well as the sociotechnical components, or as Herschel (2010, p. i) puts it “Today, the practice of BI clearly employs technology. However, it is prudent to remember that BI is also about organizational decision-making, analytics, information and knowledge management, decision flows and processes, and human interaction.” Today a plethora of definitions exist, but “there is no universally accepted definition of BI” (Wixom and Watson 2010, p. 14). In the context of this dissertation, a holistic definition of BI is adhered to following the under-standing of Wixom and Watson (2010, p. 14) who define BI as a “broad category of technologies, applications, and processes for gathering, storing, accessing, and analyz-ing data to help its users make better decisions.” The comprehension of BI in this dis-sertation includes the technological aspects with a data warehouse (DWH), the inter-faces to operational data sources, and analytical front-end applications. Further, it in-cludes the technological process view “getting data in […] and getting data out” (Wixom and Watson 2010, p. 14) as well as the organizational process view, including the soci-otechnical perspective.

Part A: Conceptual Foundations 11

Pursuant to the above presented understanding of BI, not only costs of the technological components of BI are subject to the CA for BI, but the costs of the entire BI stack (e.g., DWH, data processing, and data analysis) and also the costs incurred in the execution of related processes and the sociotechnical interaction. Therefore, a BI CA should not only consider hardware and software costs, but the full costs for BI, including, e.g., labor costs and external and internal services for BI. However, BI costs appear to be a special subject for CAs, due to BI’s nature. Subsequently, the particularities of BI in regard to CAs are briefly described:

• the usefulness of BI systems is contingent on the decisions that are taken based on information obtained from the BI system. Thus, a potential downstream value is inherent in BI systems, rather than a “value in use” that arises by using the system (Benbasat and Zmud 2003, pp. 186-190). Therefore, pricing the value of BI for CAs is difficult because neither the value of the “product” (the obtained information) nor the production costs can be easily identified.

• according to Ross et al. (1999, p. 232), the untapped potential of effective CAs is “to educate business units about IT while teaching the IT unit about the business”. This prospective benefit is particularly important for BI, since BI needs to con-tinuously meet the ever altering information requirements of the business units, e.g., in ad hoc reporting requests or data model adaptions. In contrast, business units need to expand their understanding of available BI capabilities or the system set-up itself (e.g., the underlying data models) for a purposeful use of BI systems. Thus, effective CAs can contribute “to the mutual understanding between IT and business units to exploit BI’s potential to a better extent” (Epple 2016, p. 3).

• costs of BI are mainly overhead costs (Negash 2004, p. 185) that remain as fixed costs with the BI providing unit, this is why it is a complex task to create cost transparency, to uncover inefficiencies in the use of resources, and to achieve the desired steering effects for BI resources. On one hand, this may cause a free rider problem because BI is recognized as a free good or service by the organization, which contradicts the principle of economic use of resources. On the other hand, it becomes more difficult to finance BI operations and BI investments since or-ganizational units hesitate to finance BI with their budgets, but surmise that re-quired BI systems will be financed by other organizational units.

• contrary to other domains of CAs, BI use is voluntary. Thus, BI CAs shall not restrict the usage of BI resources, but encourage users to use BI to a purposeful


extent, although BI CAs bear the potential to regulate the use of BI. This inheres a potential dilemma for BI CAs because BI CAs should not distract users from BI use; at the same time, however, BI CAs should only promote an economically meaningful extent of BI use and prevent “over-analyzing.” For IS costs, often simple usage-based CAs are implemented (Ross et al. 1999, p. 227), which might not be purposeful in the case of BI because they can result in a “death spiral” of decreasing utilization.

2.2 Cost Allocations A CA refers to the concept of internally charging “costs for the internal consumption of goods or services from one organizational unit (e.g., a cost center) to another organiza-tional unit or to a cost object (e.g., a production order)” (Epple 2016, p. 3). The under-lying rationale is that consumers should “pay” the price for receiving internally provided goods or services, and the provider should “earn” for the supply of goods or services. Therefore, a CA credits the provider and debits the consumer with a certain amount of costs. The costs credited and debited in the course of a CA are also known as secondary costs. A multitude of synonyms are used for CAs, such as chargebacks, overhead allo-cations, or internal cost/service/activity allocations. This dissertation exclusively sticks to the term CAs for consistency reasons. CAs are supposed to inhere various prospects, e.g., to enhance cost transparency, provide the correct basis for calculation, create cost awareness, or uncover inefficiencies in the use of resources (Epple et al. 2015, p. 1; Klesse 2007, pp. 37-38; Verner et al. 1996, p. 102).

A CA mechanism represents the underlying logic of how costs are transferred from sender(s) to receiver(s). The introduction of different basic CA mechanisms in this sec-tion is important because they are subject to the configuration of BI CAs. The adaptation of a basic CA mechanism for the application to specific costs (in the case of this disser-tation: BI) is regarded to as the configuration.

Figure 3 visualizes the allocation logic of basic CA that are subsequently described in detail.


Figure 3: Basic Cost Allocation Mechanisms

1. No cost allocation (Verner et al. 1996, p. 104)

BI costs are not allocated further, but remain as overhead costs with the BI-providing unit. In this case, only primary postings take place and the BI-providing cost center is not credited for providing BI to the BI users/consumers. Obviously, no CA is the sim-plest mechanism and provides no level of detail about costs.

2. Overhead rates (Verner et al. 1996, p. 104)

All primary postings are made to the BI-providing cost center. Afterwards, secondary postings distribute costs based on a certain key. The key (e.g., number of users per de-partment) is used as a rate to distribute the costs for BI to the users. The key could be based on, e.g., either actual values or on planned figures. Similarly, the cost basis to be allocated is usually based either on actual or planned costs. A commonly used synonym for overhead rates is assessment. To sum up, overhead rates are considered to be a rather rough CA mechanism that requires no high sophistication in its realization or under-standing. The provided level of detail depends on the definition of the allocation key, but is, in most cases, supposed to be lower than in other CA mechanisms.

3. Internal activity allocation (Verner et al. 1996, pp. 104)

Applying this mechanism, tariffs (prices) for predefined BI services are calculated. For the calculation of service prices, actual, planned, or market costs are used usually. The


users get debited with price multiplied by the used quantity and the provider gets cred-ited with the same amount. Synonyms are, e.g., billing or (internal) pricing. An internal activity allocation is considered a rather sophisticated CA mechanism because its im-plementation is more complex than, e.g., the implementation of overhead rates. The def-inition of activity types and tariffs adds complexity to its implementation; at the same time, however, it has the potential to significantly increase the level of provided details.

4. Activity-based costing (ABC) (Kaplan and Cooper 1998, pp. 79-110)

In ABC, the process costs as well as the cost drivers for the processes (e.g., creation of an ad hoc report) are defined. Upon execution of the process, the users are debited with the costs. The calculation of process costs could be based on planned or actual costs. The implementation and operation of ABC is sophisticated in regard to its complexity. In terms of the provided level of detail, ABC is comparable to an internal activity allo-cation.

It is noteworthy that in the configuration of BI CAs, according to the above described mechanisms, further considerations regarding the settings of single configuration com-ponents need to be made. This includes, e.g., the identification of BI services to be allo-cated, the decision on which basis the BI CA is executed (e.g., on planned, actual, or market values), or the definition of sender-receiver relationships. In the course of this decision, it is also crucial to define which part of the (BI) costs should be allocated to internal consumers, i.e., full costs or only variable costs. Further, it is important to ad-here that in one organization several CA mechanisms can be employed in different con-figurations, and even within the single domain of BI different CA mechanisms can be combined for different purposes, e.g., overhead rates for allocating the costs of BI infra-structure and internal activity allocations for allocating the costs of internal BI consul-tancy services (e.g., report adaptions).

To sum up, the basic CA mechanisms are archetypes, which need further adaptation to be applicable to BI. Moreover, further considerations regarding the management objec-tives, the contextual factors of BI CAs, and the configurable components need to be made prior to the configuration activities.

2.3 Related Work This section briefly introduces the current state of the art regarding BI CAs. A detailed overview about related work, a synthesis derived from existing literature as well as the


applied research method to identify relevant prior work can be found in paper A (cf., pp. 47-60).

Theoretical relevance

While the theoretical relevance of CAs for IS management can be found in various pub-lications (e.g., Tallon et al. 2013; Van Grembergen and De Haes 2009; Verner et al. 1996), other sources explicitly call for further research on BI management and BI cost management (e.g., Clark Jr et al. 2007, p. 603; Schieder and Gluchowski 2011, p. 12). Schieder and Gluchowski (2011) discuss future research efforts on BI cost management in order “to bring us closer to a long sought after means to assess and compare cost and benefit aspects of BI solutions” (p. 12). Stefanov et al. (2012, p. 1) state that IS CAs are “still poorly understood” and that in practice a lack of successful CAs exists.

Practical relevance

The practical relevance of BI CAs can be found in practitioner-oriented journals (e.g., Grytz 2014), or derived from practitioner-oriented publications issued by consulting companies that deal with IT CAs in general (Deloitte 2011) and BI CAs in particular (Steria Mummert 2013). Consequently, BI is a special case for CAs and of high practical relevance; however, the topic is rather unexplored in this field of research. The literature review clearly revealed a missing basis for BI CAs and the lack of substantial ground-work generally applicable to BI CAs.

Related work on cost allocations

In Section 2.2, an overview is presented of basic CA mechanisms from the accounting literature. Theoretical groundwork on cost accounting and CAs has been published by the scientific community in the 20th century (e.g., Clark 1923; Cooper and Kaplan 1988; Kaplan 1984; Riebel 1979; Shillinglaw 1989; Vatter 1950), and especially by German-speaking researchers (cf., Ewert and Wagenhofer 2011; Schildbach 1997). However, self-evidently, current IS phenomena and particularities of today’s IS were not re-searched in that era. Therefore, from an IS perspective various dedicated works on IT/IS performance management (cf., Baumöl et al. 1999; Hamel et al. 2010) and the allocation of IS costs have evolved. Research on IT/IS CAs in general (e.g., Bär and Purtschert 2014; Gerlinger et al. 2000; Laudon and Laudon 2006; McKinnon and Kallman 1987; Rom and Rohde 2007; Ross et al. 1999; VanLengen and Morgan 1993; Verner et al. 1996) as well as on the application to different domains of IS (e.g., Brandl et al. 2007; Gerlinger et al. 2000; Hosanagar et al. 2005; Müller et al. 2011; Tang and Cheng 2005; Watson et al. 2004) can be found. Generally speaking, two common characteristics can


be observed in the several sources. On one hand, the topic of CAs is mainly applied to specific sub-domains of IS, which do not reflect the particular characteristics of BI, nor can their results be broken down to the specific problem domain of BI. On the other hand, it can be observed, that the configuration of CAs depends on the design situation – consisting of intended management objectives and prevailing contextual factors –, in which they are applied. Several authors indicate the importance of certain contextual factors influencing the purposeful configuration of CAs, e.g., IS maturity or the type of services to be allocated (e.g., McKinnon and Kallman 1987; O’Connor and Martinsons 2006; VanLengen and Morgan 1993). Thus, there is no “one size fits all” solution for BI CAs; however, they must be configurable according to individual design situations.

Part A: Summary of Results 17

3 Summary of Results

The overall results of the dissertation at hand are derived from the results of five indi-vidual research works (referred to as papers), which are provided in part B (cf., pp. 47-140). In Section 3.1, an overview of the single papers is given and the interrelations between the single papers and the RQs are introduced. In Section 3.2, the contents of the single papers are summarized and the contributions of the single papers to the overall results of the dissertation are presented.

3.1 Overview Figure 4 visualizes the overall conceptual design by taking up the interrelation of RQs presented in Figure 2. It further specifies the contributions to the RQs and relates the RQs to the single papers.

Figure 4: Overall Conceptual Design

To solve the research problem, paper B contributes several generally valid design prin-ciples for BI CAs to answer RQ 1. To answer RQ 2 and to characterize the design situ-ations for BI CAs, the management objectives are identified in papers B and C, the con-textual factors are identified in paper C and further scrutinized in paper D (BI maturity) and paper E (BI acceptance). Relevant design situations are described in paper C. Due to the fact that the characterization of design situations represents a core interest of this

18 Part A: Summary of Results

dissertation, results from several papers contribute to answer RQ 2. In respect to RQ 3, the configuration of BI CAs in archetypical design situations is presented in paper C and extended in Section 4 of the synopsis paper at hand. Paper A contributes to all RQs to a certain extent because it identifies prior work relevant for this dissertation.

Table 2 further associates the single papers to the RQs and specifies the extent to which the single papers address the RQs of this dissertation.

Table 2: How the Research Questions are Addressed

Paper A does not address a particular RQ, but motivates the research and builds a useful basis for the research work conducted in the other papers, in particular for the subsequent paper B. Paper A corresponds to the “Identify Problem and Motivate” activity of the DSR process according to Peffers et al. (2007).

Paper B fully addresses RQ 1 by contributing design principles for BI CAs. Further, RQ 2 is partially covered by paper B because it identifies relevant management objectives for BI CAs.

Paper C complements the answer to RQ 2 by identifying a comprehensive set of man-agement objectives and contextual factors affecting the purposeful configuration of BI CAs. Moreover, paper C partially covers RQ 3 by adding valuable insights on the con-figuration of BI CAs in certain design situations.

Paper D as well as paper E contribute a partial coverage to RQ 2, but add a significant contribution to the analysis of design situations by providing validated instruments to


identify the states of the important contextual BI maturity and BI acceptance prevailing in organizations.

Table 2 depicts major coverage of RQ 3 in the synopsis paper at hand, although – as a matter of course – the synopsis paper addresses all RQs to the full extent. The rationale behind this depiction is that in Section 4 (cf., pp. 25-35) of the synopsis paper at hand a further contribution on the configuration of BI CAs in relevant design situations is pre-sented, which is not part of the individual papers.

3.2 Papers of the Dissertation In the course of the development of the dissertation, the research work was partitioned into five papers to be processed in manageable sets. Subsequently, for every paper the motivation, research method, results, and contribution to this dissertation are presented.

3.2.1 Paper A: Business Intelligence is no ‘Free Lunch’: What we Al-ready Know About Cost Allocation for Business Intelligence Sys-tems – and What we Should Find out

Motivation

“A review of prior, relevant literature is an essential feature of any academic project” (Webster and Watson 2002, p. 13). Correspondent to the “Identify Problem and Moti-vate” phase of the DSR process, paper A provides a comprehensive overview about prior work on BI CAs. Although CAs for BI systems are highly relevant for practice, the field is widely unexplored in current scientific literature. In this paper, the state of the art in scientific literature is assessed.

Research method

A systematic literature review according to Rowe (2014) is conducted in three iterations and enhanced by components of the hermeneutic approach according to Boell and Cecez-Kecmanovic (2014). Predefined search terms are used during each iteration and refined after the analysis of the results to be used in the subsequent iteration. A high amount of search results stemming from the three iterations is analyzed and synthesized.

Results

The literature review yielded a total of 27 relevant publications that are considered in the literature synthesis. The main outcome is that BI CAs are not yet sufficiently covered by existing literature, but several findings provide a basis for BI CAs. In consequence,


areas of future research are derived from the obtained insights. Especially, substantial research on the foundations of BI CAs, different design situations, and the configuration components is proposed. Further, in paper A (cf., pp. 47-60) detailed elements of the three propositions are specified.

Contribution to this dissertation

This paper is not intended to answer any of the RQs of the overall dissertation, but the results are supposed to deliver insights which are useful for the other papers and to mo-tivate future research on BI CAs. However, due to the autonomous character of each paper, further reviews of prior work on specific topics are necessary. This paper creates an understanding of CA mechanisms applicable for BI CAs and their impact on BI man-agement by critically reflecting prior research in this field. The paper identifies research gaps, identifies solution components, and proposes opportunities for future research.

3.2.2 Paper B: Management Objectives and Design Principles for the Cost Allocation of Business Intelligence

Motivation

This paper is built upon the findings of the extensive literature review presented in paper A. In particular, paper B refers to the first proposition for future research presented in paper A. Therefore, this paper is intended to build foundations for BI CAs in form of design principles and the identification of management objectives for BI CAs.

Research method

The findings regarding the design principles and management objectives are based on empirical data of an exploratory focus group (EFG) according to Tremblay et al. (2010) conducted with BI specialists from five major banks. The findings of the EFG are eval-uated in a confirmative focus group (CFG) according to Tremblay et al. (2010) with a different group of BI managers and BI specialists to test the findings’ validity. The de-sign principles constitute a DSR artifact according to Gregor and Hevner (2013, p. 341)

Results

The paper presents seven management objectives, three principles of form, and three principles of function for BI CAs. Further, certain generally valid (non BI-specific) de-sign principles guide the design, implementation, and application of CAs even in other domains. Thus, a solid basis for purposefully designing BI CAs is contributed. This pa-


per adds to the knowledge base of BI CAs and informs BI managers systematically de-signing BI CAs. The design principles guide the instantiation of BI CAs in given design situations and form the groundwork of this dissertation. Further, the identification of management objectives is an essential starting point for the configuration of BI CAs.


Paper B provides valuable insights on the application of BI CAs from our focus group participants and contributes a basis for the design of BI CAs. It delivers generally valid design principles that are applicable for all BI CAs independent of the design situation. Further, paper B contributes a first set of management objectives for allocating BI costs. Therefore, paper B addresses RQ 1 and RQ 2.

3.2.3 Paper C: Contextual Factors Influencing the Purposeful Alloca-tion of Business Intelligence Costs

Motivation

The literature review in paper A uncovered that CAs are highly dependent on the spe-cific design situations, but still a lack of research on constituents of design situations exists. This paper represents the major research effort toward a comprehensive set of constituents influencing BI CAs. Paper C aims at completing the picture about design situations for BI CAs by presenting a comprehensive set of management objectives and contextual factors influencing the configuration of BI CAs.

Research method

In the first step, a systematic literature review according to Rowe (2014) is conducted to identify existing work on relevant management objectives and contextual factors. A synthesis of the identified publications is reflected against data obtained from a focus group (FG) study according to Tremblay et al. (2010). In the FG as well as in the analysis of existing literature the impact on BI CAs is derived for each of the identified factors.

Results

This paper delivers valuable insights on interdependencies that need to be considered in different design situations of BI CAs. The results of this paper enhance the knowledge base on BI cost management by providing a set of management objectives and contex-tual factors (e.g., system maturity, status of acceptance, legal structure, or degree of centralization) and their type of influence on BI CAs. In total, 12 management objectives and 16 contextual factors are identified and structured in a classification scheme in form


of a morphological box. Further, three archetypical combinations of management objec-tives and contextual factors are presented and discussed.


Paper C fully addresses RQ 2 and delivers the major contribution for the identification of design situations of BI CAs. The comprehensive classification scheme (morphologi-cal box) is the summary of all identified constituents of design situations for BI CAs. The identified characteristics of the contextual factors as well as the directions of impact of the contextual factors on BI CAs contribute to a scientifically derived guideline for configuring BI CAs. The archetypical design situations and their effects on the config-uration of BI CAs serve as a basis for Section 4 (cf., pp. 25-35).

3.2.4 Paper D: Closing the Loop: Evaluating a Measurement Instru-ment for Maturity Model Design

Motivation

In a prior research initiative, a BI maturity model (BIMM) (Raber et al. 2012) and a maturity measurement instrument (MI) (Raber et al. 2013b) were designed. The MI was designed for complementing the BIMM by quantitatively assessing the BI maturity lev-els in organizations. Since according to Peffers et al. (2007, p. 54) evaluation of the designed artifact is an essential step in a DSR project, this paper concentrates on evalu-ating the validity of the MI.

Research method

By cluster analysis of maturity assessments of 92 organizations, characteristic BI ma-turity scenarios are identified for the subsequent evaluation as well as representative case companies corresponding to the scenarios. For the evaluation of the MI, its results are compared with a qualitative maturity assessment obtained by in-depth interviews in the respective case companies. A match between the quantitative maturity measurement and the maturity levels assessed in the qualitative analysis indicates that the designed artifact (the MI) correctly assesses BI maturity.

Results

In this paper, the process as well as the product (Walls et al. 2004) of the evaluation are provided. Therefore, a procedure for evaluating the BIMM is proposed and tested. The comparison of the BIMM-based maturity assessments of three organizations with the


maturity levels assessed by qualitative analyses in the same organizations shows a close match and provides evidence that the MI correctly assesses BI maturity.


BI maturity is one of the crucial contextual factors for the purposeful configuration of BI CAs. Therefore, it is essential to be able to correctly assess the BI maturity in an organization prior to the configuration activities. Paper D delivers a validated assess-ment instrument to measure BI maturity. The results of paper D contribute an important solution component to this dissertation, which could be employed for the identification of the relevant design situation for a BI CA by specifying BI maturity.

3.2.5 Paper E: Ignored, Accepted, or Used? Identifying the Phase of Ac-ceptance of Business Intelligence Systems

Motivation

The literature review in paper A reveals that research on several factors influencing the design of (IS) CAs already exists. The status of acceptance of BI systems is considered to be a crucial contextual factor for the design of BI CAs. While various IS research contributions address questions regarding the phases of acceptance and continuous sys-tem use for IS in general (e.g., Bhattacherjee and Barfar 2011), in the specific context of BI no comprehensive work exists. Therefore, the different phases of BI acceptance need to be identified by certain antecedents.

Research method

Based on a literature review this paper contributes a comprehensive set of antecedents for identifying the phases of acceptance. The validity of these antecedents is tested and empirical evidence on how the results can be employed in practice is provided by a case study according to Yin (2009).

Results

Paper E presents and evaluates six internal (attitude, cognition, behavior, affect, beliefs, and type of motivation) and three external (BI system use, learning curve, and extent of social influence) antecedents for the identification of phases of BI acceptance. The fea-sibility of operationalization is evaluated in a case study leading to the elimination of three antecedents (beliefs, type of motivation, and extent of social influence). The results


of the research provide guidance for a purposeful management of BI systems by con-tributing a solid foundation for the identification and establishment of continuous use patterns for BI systems.


Besides BI maturity, which is assessed in paper D, the state of BI acceptance is the second crucial contextual factor that is scrutinized in detail in this dissertation. To cor-rectly assess the state of BI acceptance, paper E delivers a comprehensive set of ante-cedents and a questionnaire to identify the phase of BI acceptance within organizations. Therefore, the contribution of paper E is an important solution component for this dis-sertation. The results of paper E can be employed for the identification of the relevant design situation for a BI CA, in particular by specifying BI acceptance.

Part A: Relationships Between Design Situations and Business Intelligence Cost Allocations

25

4 Relationships Between Design Situations and Busi-ness Intelligence Cost Allocations

This section demonstrates how the results – consisting of the single solution components presented in Section 3 – can be incorporated in practice to configure purposeful BI CAs. Therefore, interrelations between the intended management objectives, the prevailing contextual factors, the overall design situation, and the preferred CA mechanism are further elaborated. To this end, first, archetypical design situations originating from pa-per C (cf., pp. 83-104) are summarized in Section 4.1 followed by propositions for the purposeful configuration of BI CAs derived in Section 4.2. Further, a procedure for the purposeful configuration of BI CAs, which is based upon the aggregated results of this dissertation, is presented in Section 4.3. Hence, the insights provided in this section stem from the results of the single papers and, above all, from paper C (cf., pp. 83-104). In addition, further insights that are not beyond those in the single papers are included in the following sections. These additional insights were not included in the single papers as they either did not fit into the theme/purpose of the single papers or were revealed after connecting the results of the single papers. As such, these additional insights cater to the broader scope of this dissertation.

The results in the following sections should not be construed as concrete advice as to how single configurations should be designed since this would result in only singular and unrepeatable solutions. Instead, the findings presented here are lifted to a higher level of abstraction in order to present certain generalized findings (Lee and Baskerville 2003). Thus, the purpose of Section 4 is, on the one hand, to demonstrate the application and the usefulness of the findings in complex design situations for BI CAs by providing valuable insights in Section 4.1. On the other hand, it provides further valuable insights on the relationships between design situations and configurations of BI CAs described in Sections 4.2 and 4.3.

4.1 Business Intelligence Cost Allocations in Archetypical De-sign Situations

In paper C (cf., pp. 83-104), a classification scheme for the design situations of BI CAs is presented as a morphological box that comprises 12 management objectives and 16 contextual factors, with 62 possible characteristics for the contextual factors. Due to the

26 Part A: Relationships Between Design Situations and Business Intelligence Cost Allocations

fact that all potential combinations of characteristics of contextual factors and manage-ment objectives would result in a very high number of potential design situations, the results are condensed and certain combinations are elaborated. This approach, which is applied in paper C, was adopted from prior works in the field of contingency theory (e.g., Gordon and Miller 1976, p. 60; Reid and Smith 2000, pp. 429-430). Therefore, in this section, the archetypical design situations from paper C are briefly taken up to demonstrate the importance of design situations on the configuration of BI CAs and to serve as a basis for the subsequent sections.

However, the archetypes do not claim for completeness since it is not feasible to grasp a complete overview of all potentially existing design situations. In contrast, the three typical compositions serve as exemplary configurations and are presented here because they represent highly differentiated and complex design situations for BI CAs in prac-tice. They are differentiated covering a wide range – one bad precedent for BI CAs (the “running blind” BI company), one mediocre example for the realization of BI CAs (the “stagnant” BI company), and a good example (the “adaptive” BI company). The cate-gorization in these three archetypes can be found in various prior works and is consid-ered sufficient in this dissertation to demonstrate the application of the obtained insights. Nevertheless, this dissertation admits that a further differentiation of the presented ar-chetypes in regard to current BI phenomena might extent the comprehensiveness and account in detail for the BI specifics of the three archetypes. In order to reflect a higher degree of BI specifics in the archetypes, the characteristics of the BI-specific contextual factors could be further scrutinized or the spectrum of characteristics of BI-specific con-textual factors could be expanded, respectively. The contextual factors as well as their characteristics have been developed in paper C by an extensive literature review and a focus group study with BI experts. However, in regard to the broad definition of BI in this dissertation (cf., Section 2.1), which aims at comprising the entire BI stack and its incurring costs, a further examination of BI-specific contextual factors might be appro-priate to broaden the coverage of the contextual factors with respect to the variety of BI.

In total, three design situations that lead to three configurations are subsequently dis-cussed. In addition to the discussion, the three design situations are depicted in tables to efficiently demonstrate how the classification scheme from paper C (morphological box) can be employed to characterize design situations in practice.

The “running blind” BI company: Table 3 summarizes the design situation of the run-ning blind BI company. It stands out that the management objectives for the BI CA are


27

not clearly defined, although various BI systems with different levels of maturity and acceptance exist. One reason for that might be the low maturity of cost accounting ca-pability and the low level of management support.

Table 3: Design Situation of the Running Blind BI Company

Therefore, only a rudimentary BI CA exists in terms of overhead rates (cf., basic CA mechanisms, pp. 12-14), which is employed as an end in itself due to a lack of cost accounting maturity. This results in a stunted and ineffective BI CA due to poor align-ment with the company’s individual design situation. In this case, due to the lack of management objectives and management support, no management impact is caused by the BI CA. Therefore, in this example, no BI CA (cf., basic CA mechanisms, pp. 12-14) should be the preferable solution as this would result in a similar effect (i.e., provides no further details on BI costs, cf., Section 2.2), require less organizational effort, and reduce the potential to distract BI users from system use.

The “stagnant” BI company: Table 4 visualizes the design situation of the stagnant BI company, which contains only two cost/performance-related management objectives. Further, it needs to be highlighted that in this archetype, low BI maturity (maturity of BI services and BI management) and low BI acceptance are coupled with medium cost accounting capabilities.

Initiate

Corporate cost accounting structure Cost center

Competition & uncertainty Less competitive

Contextual factors related to environmentContextual factor Characteristics

Maturity of cost accounting Degree of centralization Centralized for data storage Decentralized for data processing and reporting(Top) management support Low

Company size Large

Phase of acceptance of BI serviceContextual factors related to organization

Contextual factor Characteristics

Various

Contextual factors related to technologyContextual factor CharacteristicsBI maturity levelType of BI system Various

Various

Management objectivesUse-/ Resource-related Cost-/ Performance-related

No clearly defined objectives


Table 4: Design Situation of the Stagnant BI Company

Consequently, rather detailed BI CAs in terms of an internal activity allocation com-bined with overhead rates (cf., basic CA mechanisms, pp. 12-14) result in a low man-agement impact due to the lack of awareness for BI. In this example, the configurations of the BI CAs seem appropriate, but slightly over-engineered; thus, they are stagnant in terms of impact. The stagnant BI company is at the edge of advancing its BI services and BI management and, therefore, realize a higher impact from its BI CAs. However, until considerable advancements in BI management are realized, a less sophisticated BI CA should be considered sufficient, i.e., a CA by overhead rates that fulfills the man-agement objectives. Overhead rates could enable the BI department to recover its costs and provide a certain level of cost transparency for BI management at the lowest ma-turity level. As a corollary, it might be argued that the sophisticated BI CAs could also foster advancements of BI services and BI management.

The “adaptive” BI company: Table 5 characterizes the design situation of the adaptive BI company. In this company, the management objectives for the BI CA are clearly defined, mature technological (BI-related) and organizational (BI management, cost ac-counting) aspects prevail, BI services are accepted, and high management support exists.

HarmonizeSupport

Adoption Adaption

InitiateIntegrate

Corporate cost accounting structure Cost centerForm of original funding of BI system Central

Competition & uncertainty Less competitive

Contextual factors related to environmentContextual factor Characteristics

Maturity of cost accounting Degree of centralization Centralized(Top) management support Medium

Company size MediumMaturity of BI management

BI innovations SeldomPhase of acceptance of BI service

Contextual factors related to organizationContextual factor Characteristics

Standardization and automation MediumTask uncertainties/interdependencies Low Medium High

Contextual factors related to technologyContextual factor CharacteristicsBI maturity levelType of BI system


Cost recovery of BI department Enhance cost transparency


29

Table 5: Design Situation of the Adaptive BI Company

As a result, the adaptive BI company realizes various highly sophisticated BI CA con-figurations for various management purposes providing different levels of details. Inter-nal activity allocations (cf., basic CA mechanisms, pp. 12-14) with predefined service catalogs and service-level agreements are in place for mature, accepted, and repetitive BI services. Overhead rates (cf., basic CA mechanisms, pp. 12-14) with different actual and planned distribution keys are used to allocate fixed costs for BI infrastructure. The BI CAs are highly effective regarding the predefined management objectives and the “as-is” configuration is considered purposeful. High levels of maturity and acceptance as well as a high level of management support allow for sophisticated BI CAs, i.e., in-ternal activity allocations, which are applied here, or ABC, respectively, in order to pro-vide the required details about BI costs.

The three characteristic archetypes described above demonstrate how the differentiated design situations influence the configurations of BI CAs and how the classification scheme from paper C can be employed in practice. Although the above described arche-types are not representative of all possible situations, which reduces the generalizability of these findings, certain interrelations between design situations and configurations can be observed that are generalizable for BI CAs. First, in the configuration of BI CAs, the design principles developed in paper B (cf., pp. 61-81) should be considered. Second, the different prevailing constituents of the design situations have different directions of impact on the configuration of BI CAs (e.g., high maturity of cost management points toward a rather sophisticated and detailed BI CA). Third, certain dominating contextual

Optimize

OptimizeOptimize

Corporate cost accounting structure Cost centerForm of original funding of BI system

Legal or contractual obligation YesCompetition & uncertainty Highly competitive

Several fundersContextual factors related to environment

Contextual factor Characteristics

Maturity of cost accounting Degree of centralization Centralized for data storage Decentralized for reporting(Top) management support High

Company size LargeMaturity of BI management

BI innovations FrequentPhase of acceptance of BI service

Contextual factors related to organizationContextual factor Characteristics

Various, but mainly accepted BI services

Contextual factors related to technologyContextual factor CharacteristicsBI maturity levelType of BI system Various

Effective BI resource utilization Enhance cost transparency


Efficient use of resources Exploit BI’s full potential Cost-by-cause allocation BI performance evaluation


factors outweigh the other constituents of the design situations. Fourth, in the running blind BI company, no clearly defined management objectives and no awareness of the contextual factors lead to an “into-the-blue” configuration of the BI CA. Thus, it is es-sential to follow a certain procedure in the design process of BI CAs, e.g., to define the management objectives in a first step prior to the configuration of BI CAs. Conse-quently, in Sections 4.2 and 4.3, respectively, these insights are incorporated in three propositions and further reasoning on the design procedure for BI CAs is discussed.

4.2 Propositions for the Configuration of Business Intelligence Cost Allocations

Subsequent to demonstrating exemplary relations between design situations and the con-figurations of BI CAs, the obtained insights are clustered in three propositions for BI CAs. Proposition one and two are based on the results presented in the constitutive pa-pers of this dissertation, whereas proposition three reflects further insights that have not yet been published elsewhere. The further insights presented in this section complement the answer to RQ 3 and enhance the results presented in paper C.

Proposition 1: Adherence to the design principles avoids basic configuration problems of BI CAs

The configuration of BI CAs is heavily influenced by the design principles; hence they need to be developed accordingly. In the course of this dissertation, generally valid de-sign principles were developed (paper B, cf., pp. 61-81) to which every BI CA must abide. In the above described archetypes, the violation of two of the design principles can be observed, which leads to not purposeful BI CAs. In the running blind BI com-pany, the CA is employed as an end in itself, which contradicts design principle #8 (cf., paper B, pp. 72-75), which states, “a BI cost allocation should not be an end in itself, but be a means to realize certain management goals” (Epple et al. 2015, p. 9). The con-figuration in the running blind BI company exposes the configuration problem of miss-ing management objectives, thus leading to a non-purposeful BI CA, which is taken into consideration by design principle #8. In the stagnant BI company, the BI CA configura-tion is overly sophisticated in relation to the design situation. Design principle #3 (cf., paper B, pp. 72-75) takes this configuration problem into account, stating that “overly complex allocation logics need to be avoided as they are counterproductive and not eco-nomic” (Epple et al. 2015, p. 8). At this point, it is noteworthy to emphasize the eco-nomic efficiency since it is crucial for BI CAs. Other authors (e.g., Klesse 2007, p. 39;


31

Verner et al. 1996, pp. 110-111) also mention economic efficiency as one basic require-ment for CAs. This requirement asserts that the relation between costs and benefits of a CA must be considered in the configuration of a CA and why a balance must be struck between the accuracy of a BI CA and the necessary effort to implement and operate it. Potentially, a 100% accurate BI CA might be realized; however, this also probably in-volves significant effort (cost) to retrieve and analyze the data compared with the bene-fits achieved from an accurate BI CA. In such cases, the particular consideration of man-agement objectives and contextual factors may contribute to the configuration of an eco-nomic BI CA by envisioning the management objectives and contextual factors, thereby bringing the necessary efforts in relation to the expected benefits. Generally speaking, it can be observed that consideration of the design principles developed in paper B can lead to more purposeful configurations of BI CAs by avoiding basic configuration prob-lems.

Proposition 2: The consideration of directions of impact of contextual factors provides guidance for the detailed configuration of purposeful BI CAs

The results of this dissertation (cf., paper C, pp. 83-104) show that single characteristics of contextual factors have impacts in certain directions on the configuration of BI CAs. In this context, the direction of impact refers to the effect of a contextual factor charac-teristic on the configuration of a BI CA. Put differently, it describes the effect a certain contextual factor has on the configuration of BI CAs, e.g., a high maturity of BI man-agement induces a sophisticated configuration that provides details about the allocated BI costs. In the above described stagnant BI company, a sophisticated BI CA is em-ployed, although low BI maturity and low BI acceptance prevail. According to the re-sults of paper C (cf., pp. 83-104) “a sophisticated CA might not be purposeful, and might even cause effects contrary to their intended results, if BI services are not even accepted yet” (Epple 2016, p. 10). This statement on the impact of the contextual factor BI ac-ceptance proposes to direct the configuration toward a less sophisticated BI CA if low BI acceptance prevails in an organization. Therefore, on a general level, the considera-tion of the direction of impact of contextual factors is crucial for effectively configuring purposeful BI CAs. In paper C, the directions of impact of single contextual factors are presented to the reader. Therefore, in this section the direction of impact should not be repeated, but referred to as the basis for this proposition.


Proposition 3: The dominating contextual factors of a design situation are the most influencing factors for BI CA configurations

Among the configurations depicted in Section 4.1, it can be observed that certain char-acteristics of contextual factors outweigh the impacts of others; e.g., in the running blind BI company, the low level of management support makes the BI CA ineffective in terms of impact, although various levels of BI maturity exist. Consequently, the concept of dominating contextual factors is introduced. Dominating contextual factors are those whose impact on the configuration of BI CAs exceeds the impact of other contextual factors. The concept of dominating contextual factors becomes especially important in the case where one contextual factor points to a certain configuration of BI CAs while at the same time another prevailing contextual factor suggests a different direction of impact for the configuration. Table 6 provides an overview of potentially dominating contextual factors and their impacts on BI CA configuration. The subsequent results are derived from insights obtained in the analysis of the archetypical design situations in Section 4.1 as well as by critical reasoning.

The below described impacts of dominating contextual factors of design situations pro-vide guidance on how to configure BI CAs in design situations with dominating contex-tual factors. However, often no dominating contextual factor is identifiable. In contrast, a design situation with several dominating contextual factors may also occur in practice. In these cases, the overall design situation and the impacts of the single contextual fac-tors need to be analyzed and balanced; thus, a structured procedure, which is presented in the following section, shall provide a systematic approach for the necessary steps toward a purposeful configuration.


33

Table 6: Dominating Contextual Factors

4.3 Procedure for Business Intelligence Cost Allocations The exemplary configurations in Section 4.1 show that the search process for the optimal BI CA needs to follow a certain procedure to avoid non-purposeful “into-the-blue” con-figurations. Therefore, the provision of a structured procedure that incorporates the re-sults of this dissertation is essential for their effective application in practice. Guidelines on necessary steps for the configuration of CAs can be found in prior works (e.g., Verner et al. 1996, p. 114) that partially incorporate results obtained in this dissertation, e.g., the consideration of management objectives. Therefore, the systematic procedure pre-sented in this section extends prior works by incorporating the results for the design

Dominating con-textual factor

Character-istic Impact on the configuration of BI CAs

Company size Small According to Drury (2000, p. 298) IS CAs are mainly em-ployed by big companies. Therefore, it is supposed that BI CAs do not add value in small companies in most cases.

Phase of BI ac-ceptance

Initiation, Adoption, Adaption

A BI CA in an early stage of BI acceptance inheres the po-tential to distract BI users from system use. Thus, in this phase, no BI CA, an easily understandable BI CA, or a use-promoting (e.g., with negative prices) is preferable.

BI maturity level Initiate, Harmonize

The sophistication and level of detail of BI CAs supposedly should correlate with the BI maturity level. Sophisticated BI CAs for low maturity BI services are not considered ap-propriate.

Maturity of BI management and/or Maturity of cost accounting

Initiate, Harmonize

Low organizational maturity (BI management or cost ac-counting) appears to contradict sophisticated BI CAs. If both organizational maturities are on a low level, then no BI CA should be preferable. In the case of one low organiza-tional maturity, it has to be thoroughly evaluated if the other organizational capabilities can compensate the ab-sence of the counterpart.

(Top) manage-ment support Low

No or low (top) management support can hinder the pur-poseful realization of BI CAs. Missing management sup-port indicates no or simple BI CAs to be preferable.

Corporate cost accounting struc-ture

Separate company

If BI is organized in a separate company, then an internal CA (within the group of companies) should be avoided; however, a billing process via financial accounting has to be realized.

Legal or contrac-tual obligations Yes

If no other management objective or contextual factor calls for a more sophisticated BI CA, then the configuration could be targeted to fulfill only the minimum requirements of the obligation.


principles, management objectives, contextual factors, the directions of impact of con-textual factors, and conception of dominating constituent in a systematic way. To con-figure a purposeful BI CA that is appropriate for a given design situation, it is important to adhere to the presented procedure.

Figure 5 presents a procedural flow chart for the design, implementation, and continuous revision of BI CAs. Prior to starting configuration activities, the design situation needs to be identified and analyzed. Therefore, in the first step, the management objectives of the BI CA need to be ascertained to determine the desired management impacts that shall be achieved. In the second step, the characteristics of prevailing contextual factors need to be analyzed. In the course of the second step, a further detailed assessment and analysis of the important contextual factors of BI acceptance and BI maturity might be conducted to determine the particular influence of these factors. After the identification of all relevant constituents of the design situation, the design situation needs to be clas-sified and the impact of the design situation on the configuration analyzed considering the above presented propositions (cf., Section 4.2). Following this analysis, an appro-priate CA mechanism can be selected and the actual configuration of the BI CA takes place (e.g., definition of allocated BI services, definition of cost types to be allocated). In the fifth step, the BI CA is applied over a certain period of time. It is crucial to adhere to the BI CA and grant enough settling time to the BI CA because the desired manage-ment effects will not become immediately effective. For the success of the BI CA, those involved as stakeholders (e.g., BI provider and BI consumers) need to gather experience with the BI CA, become acquainted with the new measure, and establish confidence and acceptance for the BI CA. Nevertheless, a continuous evaluation and (if necessary) re-finement of the BI CA needs to take place, which is represented by the sixth step. In the evaluation, on one hand, the achievement of the desired management effects needs to be assessed. On the other hand, the design situation needs to be repeatedly analyzed since steps one to three always represent a snapshot at a certain time and management objec-tives as well as contextual factors tend not to be stable but continuously developing. Therefore, over time, various design situations prevail even in a single company, which is why continuous revisions and updates of BI CA configurations need to be conducted.


35

Figure 5: Procedure for the Configuration of Business Intelligence Cost Allocations

This procedure places an emphasis on the importance of design situations on the con-figuration as well as on the revision of BI CAs. However, it also points out the particular influence of certain contextual factors in the context of BI, in particular of BI acceptance and BI maturity, which need to be considered as central to BI CAs. Depending on those two central contextual factors, BI-specific benefits, such as fostering BI use to a pur-poseful extent and BI-/business-alignment, could be achieved.

36 Part A: Evaluation

5 Evaluation

The objective of DSR is to produce prescriptive knowledge in the form of artifacts that help to solve design problems in the real world. In contrast to other research disciplines (natural or human sciences), the evaluation of research results should not only be con-ducted ex post, but ex ante iteratively during the design process (Abraham et al. 2014; Sonnenberg and vom Brocke 2012; Venable et al. 2012). Due to the cumulative nature of this dissertation, the individual papers partially constitute independently coherent DSR projects that pass through all or dedicated steps of DSR projects. In the single papers, certain evaluation steps have been conducted to evaluate the solution compo-nents ex ante. In Section 5.1, the evaluations of solution components that have taken place in the single papers are described. Section 5.2 depicts an approach for the overall evaluation of the results of this dissertation.

5.1 Evaluation of Solution Components Except for paper A, in all single papers certain evaluation steps have been conducted.

In paper B, an exploratory focus group study has led to the construction of a DSR artifact in the form of design principles and also to the identification of management objectives for BI CAs. FGs can not only be employed for exploratory purposes, but according to Sonnenberg and vom Brocke (2012, pp. 393-394), they are appropriate for every evalu-ation step of the design process. Therefore, a CFG was conducted to validate the findings from the EFG. By analyzing the results of the CFG, certain results of the EFG had to be dismissed, which led to their refinement and confirmation.

In paper C, a similar approach to paper B is employed, but with a single FG study for exploratory and confirmatory purposes. In paper C, the synthesis of an extensive litera-ture review led to certain results that were extended and confirmed through the FG. However, the confirmatory aspects of the FG played a subordinate role and are not re-ported in detail in paper C since the main goal of the FG was to extend the results of the literature review.

Evaluation is the crux of the matter in paper D as indicated by its title. Paper D builds upon prior research, in which a BI maturity model and a respective measurement instru-ment were designed. In paper D, an evaluation strategy is devised and the validity of the MI is tested. In a first step, to identify cases for evaluation, four typical BI maturity scenarios were derived through cluster analysis. In a second step of the evaluation, in-

Part A: Evaluation 37

depth interviews were conducted with the identified case companies to qualitatively as-sess their BI maturity. In the last step of the evaluation, the results of the qualitative assessment were compared with the quantitatively obtained results from the BI maturity MI. A close match between the qualitative and quantitative assessments indicates the validity of the BI maturity MI. Moreover, the evaluation approach developed in paper D has the potential to be reused in further research.

In paper E, first the antecedents for identifying phases of BI acceptance are derived from a comprehensive review of prior work. In the second step, the identified antecedents are validated and tested for practicability in a case study. Therefore, the evaluation of the results obtained in the literature review in a practical case is one of the key contributions of paper E.

In conclusion, the evaluation of results is a central component in four of the five papers. Thus, the single solution components have already been evaluated in the course of the single research paper projects. However, since the overall contribution is derived in the synopsis paper at hand, the evaluation of the overall contribution is pending and is dis-cussed in the subsequent section.

5.2 Overall Evaluation To complete a DSR project according to the six-step approach proposed by Peffers et al. (2007), an evaluation of the overall contribution should be conducted (ex post eval-uation according to Sonnenberg and vom Brocke (2012)). In the following, first con-ceivable subjects for further evaluation are introduced. Subsequently, reasoning is pro-vided as to why an evaluation of the overall contribution is not feasible in this disserta-tion in a straightforward process from conceptual design over implementation until the overall evaluation. Concluding, the evaluation strategy devised in paper D is taken up in order to develop a realistic evaluation concept for the results of this dissertation.

The presented results can be further evaluated by certain means. Therefore, the follow-ing three notions are presented that could stimulate future evaluations of this disserta-tion’s results:

1. Although the directions of impact of the single contextual factors on the config-uration of BI CAs have been part of the evaluation in the FG in paper C (cf., pp. 83-104), they have not been the focal point of the evaluation. Therefore, the re-lations between design situations and the configuration of BI CAs as well as the propositions derived in Section 4.2 could be subject to further evaluation.

38 Part A: Evaluation

2. The proposed procedure for the configuration of BI CAs (cf., Section 4.3) could be evaluated. Due to the fact that the procedure is derived in part A, it has not yet been evaluated in one of the single papers. Therefore, the procedure shall be fur-ther evaluated to test its validity and extend it using insights from practice.

3. The usefulness of the aggregated results of this dissertation, i.e., the design prin-ciples, the set of instruments to identify and analyze the design situation, the propositions, and the proposed procedure, could be subject to an overall evalua-tion. Thus, an extensive and integrated evaluation strategy would have to be set up to comprehend the entire set of results.

Notions 1. and 2. refer single solution components introduced in the synopsis paper at hand, whereas notion 3. is directed towards an integrated evaluation approach for the aggregated results of this dissertation.

An evaluation of the overall contribution in a straightforward process is not feasible in the course of this dissertation mainly for the following three reasons. First, to obtain appropriate data for an evaluation, either an existing BI CA in an organization would have to be changed following the rationales of this dissertation, or a BI CA would have to be set up with a “greenfield” approach in accordance to the results. Typically, changes of this extent are organized in larger transformation projects, which goes beyond the scope of this dissertation since access to such transformation projects dealing with sen-sitive cost data is, in most organizations, heavily restricted and not an open field for researchers for evaluation purposes. Second, if an evaluation of the overall contribution would be possible in a certain organization, it would take place for exactly one design situation: the design situation of the respective organization. Consequently, the evalua-tion would have to be conducted in multiple cases to achieve sufficient data to ensure the evaluation’s reliability and validity. Third, even in the case where one or several evaluations can be conducted, the results would have to be evaluated in a longitudinal study because the goodness of a BI CA that follows the results can only be observed retrospective to implementing the changed/newly configured BI CA, and not immedi-ately afterwards. Laying out an evaluation on a longitudinal or repetitive basis, respec-tively, is not feasible in the course of this dissertation due to the limited scope and timeframe.

Although the obstacles for evaluating the overall contribution in a straightforward pro-cess comprising design, implementation and evaluation are introduced above, a realistic evaluation concept shall be discussed in the following. Due to the complexity of the subject and the difficulty to obtain mass data regarding CAs, likely, a feasible evaluation

Part A: Evaluation 39

approach is qualitative by nature. Further, following the underlying concept of the qual-itative evaluation strategy developed in paper D a retrograde evaluation approach could assist to overcome the difficulties of evaluating the overall results in a linear straight-forward process. In paper D a qualitative assessment of maturity levels confirmed the validity of the developed BI maturity MI. Transferring the underlying concept of this evaluation strategy to the overall results of this dissertation implies a viable evaluation concept in two steps. In the first step, the design situations, the configurations of BI CAs as well as the attainment of management objectives and the perceived reasonableness of the applied BI CAs have to be assessed in different case companies. Following the eval-uation approaches of paper D and E, in-depth interviews or case studies could be the appropriate research method for the first step. In a second step, the configurations of BI CAs in those cases with a high attainment of management objective and high perceived reasonableness could be compared to the actual configurations of BI CAs that would have been indicated by the results of this dissertation in the design situations of the re-spective case companies. A close match with no significant deviations in the two con-figurations of BI CAs indicates the validity of this dissertation’s results. Further, the obtained insights from the real-world evaluations in case companies could have the po-tential to enhance the results of this dissertation. In contrast to a straightforward evalu-ation approach preceded by the conceptual design and implementation of a BI CA ac-cording to the results of this dissertation, the proposed retrograde evaluation concept bears a high practical feasibility paired with the potential to have feedback loops for the extension of the results.

40 Part A: Discussion

6 Discussion

In this section, the results of this dissertation as well as its limitations are reflected. Fur-ther, the implications for research and practice are critically discussed and directions for future research presented.

6.1 Summary and Limitations The overall research objective is to provide prescriptive knowledge as a set of instru-ments that simplifies, systematizes, and guides the quest for optimal and purposeful BI CAs. Further, the research objective needs to be achieved under the particular require-ment of incorporating the specific design situations in which BI CAs are applied. This research objective was operationalized through the three RQs presented in Section 1.3. RQ 1 and RQ 2 target the understanding of the problem domain, whereas RQ 3 aims at designing certain generalizable solution approaches. The answers to the RQs and the overall results are an aggregation of results from the single papers, which are based on rigorous research methods. In the synopsis paper at hand, the results are put in context and enhanced using additional findings from the dissertation project that have not been published in the individual papers. The main contribution is the advancement of research on BI CAs by contributing a suite of useful instruments for the situational configurations of BI CAs.

This research is motivated by the understanding that no “one size fits all” approach can be developed for the configuration of BI CAs, but that the specific design situations play a decisive role for the configuration of BI CAs. Therefore, this research first reveals design principles that are valid for all design situations of BI CAs and that provide a foundation for configuring BI CAs (RQ 1). In the second step (RQ 2), a classification scheme for the design situations is developed and instruments used to assess the im-portant contextual factors BI maturity and BI acceptance. In this context, BI maturity and BI acceptance are supposed to have a high influence on the purposeful configuration of BI CAs, thereby justifying the detailed analysis undertaken in this dissertation. In the third step (RQ 3), the relationship between design situations and the configuration of BI CAs is demonstrated and certain generalized propositions are derived. Moreover, a pro-cedure for developing purposeful configurations is presented. Table 7 gives a brief over-view of the contributions from the single papers in relation to the RQs.

Part A: Discussion 41

Table 7: Contributions Related to Research Questions and Papers

RQ # Research Questions and Contributions Paper RQ 1 What are the design principles for BI cost allocations?

Establishment of a basic understanding of the research problem and possible solution approaches, i.e., the need for foundations of BI CAs in the form of design principles. Delivers the basis for the develop-ment of design principles in paper B.

A

Development of six BI-specific design principles for BI CAs and three generally valid design principles for (IS) CAs.

B

RQ 2 What characterizes relevant design situations for BI cost allocations? Understanding how design situations for BI CAs are constituted. Identification of prior work relevant for further characterization of de-sign situations.

A

First identification and understanding of management objectives for BI CAs in an EFG and evaluation in a CFG.

B

Delivery of a comprehensive set of management objectives and con-textual factors to understand design situations of BI CAs. Develop-ment of a classification scheme to identify and characterize design sit-uations of BI CAs.

C

Validated assessment instrument to assess BI maturity, which is considered a crucial contextual factor. Contributes an important solu-tion component for the identification of relevant design situations.

D

Analysis of the second crucial contextual factor: BI acceptance. Deliv-ery of a set of antecedents as well as a questionnaire to correctly identify the phase of BI acceptance in design situations for BI CAs.

E

RQ 3 How should the allocation of BI costs be configured in a certain design situa-tion? Understanding of the relationships between design situations and configurations of BI CAs.

A

Derivation of directions of impact of all identified contextual fac-tors on the configuration of BI CAs. Identification of archetypical de-sign situations and provision of reasoning for their particular BI CA configurations.

C

Discussion on the impact of the respective design situation on the con-figuration of BI CAs. Derivation of three generalized propositions by reflecting the results of the single papers. Introduction of the concep-tion of dominating contextual factors and presenting their impacts. Provision of a procedure for the configuration and continuous revi-sion of BI CAs, which incorporates the results of this dissertation.

Summary paper

Dissenting from the “one size fits all” approach, this dissertation, on one hand, adds clarity to the topic of BI CAs from a research perspective by delivering a set of instru-ments applicable in different design situations for BI CAs. However, on the other hand, it must be acknowledged that the examination of the topic by means of research in prac-tice might appear overly complex in certain design situations or might add unnecessary complexity to finding an optimal BI CA in certain situations. In Section 4.2 the basic


requirement of economic efficiency is briefly discussed stating that a balance between the costs and benefits of a BI CA must be struck. Consequently, the “one size fits all” hypothesis might be valid for certain design situations, in which the additional efforts exceed the added value of a BI CA. In such cases not only directly quantifiable costs shall be regarded as additional efforts, but factors like missing acceptance, implementa-tion restrictions, or alignment with the corporate cost accounting logic must be consid-ered in the comparison with the added value. Therefore, this dissertation admits that a “one size fits all” solution or at least a to a certain extent standardized solution could be preferable for certain design situations, in which the application of this dissertation’s results is not economically reasonable. Further, irrespective of economic considerations, in certain design situations a “one size fits all” solution might directly be the ideal solu-tion for a BI CA. Table 6 provides an overview of dominating contextual factors indi-cating that certain characteristics of contextual factors imply only dedicated configura-tions of BI CAs. Thus, a “one size fits all” solution could be directly applicable if certain characteristics of contextual factors prevail, e.g., in small companies with no manage-ment support and low management accounting capabilities the preferable “one size fits all” solution could always be no BI CA irrespective of further characteristics of the in-dividual companies. Due to the high amount of potential design situations a variety of situations, in which one uniform configuration of BI CA is the optimal solution seems plausible. Hence, the dominating contextual factors presented in Section 4.2 could serve as a basis to derive further characteristics of contextual factors that directly suggest a certain “one size fits all” solution.

No research is without limitations. Therefore, this dissertation needs to be reflected upon given its limitations.

First, as described in Section 5.2, an overall evaluation of the aggregated results has not been conducted due to the above mentioned reasons. The missing overall evaluation might be to the detriment of the overall validity of the aggregated results. Nevertheless, the results of the single papers have been subject to evaluations, why individually seen the components are already evaluated. However, an overall evaluation could enhance the validity and lead to a refinement of the results. Thus, evaluation strategies that lie beyond the scope of this dissertation are proposed in Section 5.2.

Second, this dissertation does not provide a panacea for the configuration of BI CAs, rather, it has broken down a real-world design problem into manageable sets; thus, the results do not give concrete advice how the BI CAs must be designed in every possible design situation. Although it is recognized that not every potential design situation can


be researched, certain general patterns and relations are observed. Therefore, the results are kept on a higher level of abstraction in order to be generalizable to some extent.

Third, the three propositions as well as the procedure presented in Section 4 are only argumentatively derived in the synopsis paper at hand based on the results of the single papers. The results presented in Section 4 might lack of scientific rigor and have not yet been subject to a thorough review process in an IS journal or IS conference. Therefore, the results of Section 4 need to be considered as work-in-progress to some extent and might be subject for future research. Nevertheless, the results have been derived to the best of the author’s knowledge striving for meticulous derivation and presentation.

Fourth, the results of the single papers are achieved by qualitative research, i.e., FG studies in papers B and C, in-depth interviews in paper D, and a case study in paper E. Although the applied research methods are considered appropriate in the single papers, owing to the nature of qualitative research, biases might have slipped in. To the best of the researcher’s knowledge, countermeasures have been taken into account, e.g., the documentation and the subsequent consolidation of results by several researchers to avoid misinterpretations (Miles and Huberman 1994, p. 64). Nevertheless, as inherent to qualitative research potential misinterpretations cannot be completely avoided.

6.2 Implications for Practice and Research According to Winter, DSR needs to be scientifically rigorous and practically relevant (Winter 2008, p. 470).

From a practical perspective, the research topic is highly relevant, which becomes evi-dent from publications in practitioner-oriented outlets (cf., Section 2.3) as well as from the FG studies conducted in the course of the single paper projects. Therefore, the results are thought to have a high impact for practice; concurrently, the results are expected to be generalizable from a research perspective. This balancing act was accomplished by providing a set of scientifically grounded instruments to practice. More specifically, this dissertation has the following implications for practice.

First, the results create awareness for the specifics of BI CAs, which need to be consid-ered during their development. Therefore, for the configuration of BI CAs directly ap-plicable design principles are contributed. Second, the comprehensive classification scheme assists practitioners in the identification of their design situations. The classifi-cation scheme, as well as the instruments to assess BI maturity and BI acceptance, serves as a practical orientation in the search process for BI CAs. In this way, the process gains


efficiency and may lead to the desired results. Third, the presented relations between design situations and the configuration of BI CAs with the derived propositions provide a catalog of implications that need to be considered by practitioners striving for a pur-poseful BI CAs. Moreover, the presented procedure gives specific guidance on neces-sary activities to avoid non-purposeful “into-the-blue” configurations.

Practitioners are provided appropriate instruments to achieve the desired impacts of BI CAs. In accordance with the research results, business units could, e.g., come closer to BI departments, understand incurred BI costs, create the necessary transparency to en-hance BI services, and thereby exploit the prospects of BI systems to the best of its capabilities.

The implications for research are as numerous as the implications for practice.

First, this dissertation sheds light on the identified research gap and answers to several calls for future research on BI cost management issues. Further, it contributes to the knowledge base of BI CAs and BI cost management by providing a comprehensive study of BI CAs. On the aggregated level, several artifacts are contributed to the knowledge base that represent a valuable collection of prescriptive knowledge. Further, the results are put in context in the work at hand. Second, on the level of the individual papers, every paper is a self-contained research work that contributes results to its iden-tified research gap by undergoing a thorough research process. Thus, the implications of this dissertation for research can be considered either on an aggregated or individual-paper level. Looking forward, the implications of the individual papers can be found in the single papers; however, several connecting points to future research remain, which are presented in Section 6.3.

6.3 Fields for Future Research “Science is a cumulative endeavour as new knowledge is often created in the process of interpreting and combining existing knowledge” (vom Brocke et al. 2009, p. 2206). Put differently, the fruitfulness for future research (Aier and Fischer 2011, pp. 155-156) is an important criterion for research works. Therefore, in the single papers of this disser-tation, the corresponding fields for future research are presented. On the aggregated level of the results, the fields for future research are listed and briefly discussed below:

- Conduction of proposed evaluation activities: in Section 5.2, different notions to approach an overall evaluation of the results of this dissertation are presented. Pursuing those notions could lead to additional insights and an extensions of the


results. In this context, an instrument to assess the goodness of a BI CA in respect to its fit to the individual design situation would be desirable.

- Carrying the results of Section 4.2 forward: as described above, the results pre-sented in Section 4.2 have been partially derived in the synopsis paper at hand. Therefore, these results bear a high potential to be extended. Links to future re-search might be the proposed evaluation activities as well as the extension of the results by analyzing further data.

- Development of an assessment instrument to identify the design situations: the results of this dissertation assist the identification of respective design situations for BI CAs. However, the development of an integrated assessment instrument that leads through the identification process of design situations would comple-ment the overall results. Content-wise, the components of the corresponding as-sessment instrument are provided in this dissertation to a high extent. Besides the assessment instruments for the complex contextual factors BI acceptance and BI maturity, the assessment instrument would have to incorporate measurement in-struments for the non-evident characteristics of contextual factors.

- Complementation by other measures: a BI CA is not a panacea to achieve all the desired management impacts in a stand-alone manner, but it is one central meas-ure of cost management. For decision-making based on the results of BI CAs, other measures need to complement the results of BI CAs. For example, to eval-uate the performance of a BI department, complementary performance indicators need to be calculated; or, to evaluate outsourcing decisions, comparability of costs needs to be created. Therefore, a set of complementary measures – depend-ing on the management objectives – would enrich the results of this dissertation and boost its practical impact.

Part B: Business Intelligence is no ‘Free Lunch’: What we Already Know About Cost Allocation for Business Intelligence Systems – and What we Should Find out

47

Part B – Papers of the Dissertation

Paper A – Business Intelligence is no ‘Free Lunch’: What we Already Know About Cost Allocation for Business Intelligence Systems – and What we Should Find out

Table 8: Bibliographical Information for Paper A

Title Business Intelligence is no ‘Free Lunch’: What we Already Know About Cost Allocation for Business Intelligence Sys-tems – and What we Should Find out

Authors & Affiliation Johannes Epple, Stefan Bischoff, Stephan Aier, Robert Winter

University of St. Gallen, Institute of Information Management, Mueller-Friedberg-Strasse 8, 9000 St. Gallen, Switzerland [email protected], [email protected], [email protected], [email protected]

Publication Outlet Institute of Information Management, University of St. Gallen, St. Gallen 2016

URL https://www.alexandria.unisg.ch/publications/249525

Publication Type Working Paper

Publication Year 2016

Publication Status Published

Rating (VHB1 Jourqual 3) -

Abstract

Cost allocations for business intelligence (BI) systems are supposed to create cost awareness, enhance cost transparency, and support the management of BI systems. Alt-hough cost allocations for BI systems are highly relevant for practice, yet the field is widely unchartered in current scientific literature. In our paper we assess the state of the art of this field in scientific literature. We describe three literature review iterations. While several approaches exist for allocating costs in the field of information systems,

1 (Verband der Hochschullehrer für Betriebswirtschaft 2015)

48 Part B: Business Intelligence is no ‘Free Lunch’: What we Already Know About Cost Allocation for Business Intelligence Systems – and What we Should Find out

no publication particularly focusses on the allocation of BI costs, but various sources deliver components that can be used for the design of a BI cost allocation. We synthesize the results obtained in the literature review and derive an agenda for future research.

Keywords

IS Management, BI Governance, Cost Accounting, Business Intelligence.

A.1 Introduction Today, Business Intelligence (BI) is widely understood as an umbrella term for “tech-nologies, applications and processes for gathering, storing, accessing and analyzing data to help its users make better decisions” (Wixom and Watson 2010, p. 14). While general information technology expenses only rose by 0.4% in 2013 (Gartner Inc. 2014a), ex-penses for BI grew by 8% in 2013 (Gartner Inc. 2014b). BI is no longer a “nice-to-have” report generator; it is considered a prerequisite for organizational success (Wixom and Watson 2010). Still, there is a high need within organizations to justify the rather high and growing expenses for BI. However, the costs of BI are mainly overhead costs (Negash 2004). It is, therefore, difficult to create cost transparency and cost awareness, to uncover inefficiencies, and to generate desired steering effects for the management of BI resources that are used by other units (Olson and Ives 1982; Verner et al. 1996).

Cost accounting and cost allocation (CA) have been addressed by management research for a long time (e.g., Clark 1923; Cooper and Kaplan 1988; Shillinglaw 1989) in the dominant context of the manufacturing industry, where overheads have to be allocated to produced goods. Information systems (IS) instead have more infrastructure charac-teristics (creating potentials, shared resources) than manufacturing (creating non-shara-ble outputs from input goods). According to Deloitte (2011), the maturity level of IS CA in practice is still low. Stefanov et al. (2012) confirm that IS CAs are “still poorly un-derstood” and that there is a lack of successful CAs in practice. Publications crossing the line between IS and management accounting exist (e.g., Rom and Rohde 2007), but do not consider the particularities of BI. Owing to BI’s particular characteristics, BI seems to be of special interest to cost accounting. First, the output of BI is data converted into information that is difficult to price, because neither the production costs nor the value of the obtained information can be straightforwardly identified. Second, the ben-efit of a BI system lies in the purposeful use of the system’s output rather than in the BI system use itself (Benbasat and Zmud 2003). Third, BI investments and operation create a high monolithic cost block (Bischoff et al. 2014) that remains unmanageable without


49

allocating the cost to the causer. Fourth, for IS often pay-per-use CAs (Ross et al. 1999) are implemented that restrict the use of resources, which is not purposeful for BI. In contrast, a BI CA shall not distract from BI system use, but promote BI system use to the best of its capabilities to reach beneficial business decisions. Several publications particularly call for research on cost issues of BI (Arnott and Pervan 2008; Clark Jr et al. 2007; Schieder and Gluchowski 2011). Therefore, in this paper, we aim to delineate the playing field and to analyze prior research on CA of BI systems. Consequently, this paper addresses the following research questions (RQ):

1. What are the existing methods for cost allocation of BI systems? 2. What are the fields for future research regarding BI cost allocations?

Our paper establishes an understanding of CA methods applicable for BI systems. We identify research gaps and propose opportunities for future research. This paper is orga-nized as follows: section two presents the conceptual foundations important for deriving the search terms. Section three introduces the applied research method. Sections four and five present the iterative literature review and the synthesis of the results, respec-tively. In section six, we summarize the findings and discuss our results.

A.2 Conceptual Foundations In this section, we introduce CA methods commonly recognized in accounting literature. Furthermore, a terminology for BI is established that also clarifies similarly used terms. The result is the identification of main search terms for the literature review as recom-mended by vom Brocke et al. (2009). A CA method relates to the underlying allocation mechanism that transfers costs from provider to consumer. The differentiation of cost accounting methods is crucial for the literature review because the different denomina-tions need to be considered as search terms. The following general CA methods exist:

• No cost allocation (Olson and Ives 1982): the costs are not further allocated, but remain with the providing unit as overheads.

• Overhead rates (Verner et al. 1996): a key (e.g., number of users, CPU usage) is employed as a ratio according to which the BI costs are distributed to the con-sumers. Synonym: assessment.

• Internal activity allocation (Verner et al. 1996): prices for activities are defined. The consumers are debited with price * used quantity. Synonyms: billing, internal cost allocation, (internal) pricing, charging, chargeout, and chargeback.


• Activity based costing (ABC) (Kaplan and Cooper 1998): costs are assigned to processes and process cost drivers (e.g., creation of a report) are determined. Upon performing a cost driver, the costs are debited to the consumers.

• Relative direct cost calculation (Ewert and Wagenhofer 2011): a rather special CA method that assigns costs to cost objects related to the business decisions that cause the costs.

These methods describe the basic mechanisms of how costs are allocated from provider to consumer and are archetypes of mechanisms needing further purposive adaptation in order to be applicable to BI systems. Restricting the literature review to the term BI can lead to the exclusion of relevant results. The term data warehouse (DWH) is often men-tioned in the same breath with BI although it describes an enabling component for BI (Ramamurthy et al. 2008). The term analytics is often used synonymously for BI (Chen et al. 2012). Consequently, the terms “data warehouse”, “data warehousing”, and “ana-lytics” are incorporated into the literature review.

A.3 Types of literature reviews Literature reviews represent an important contribution to research (vom Brocke et al. 2009) and are recognized as “an essential feature of any academic project” (Webster and Watson 2002, p. 13). Different types of literature reviews exist, among them the so-called systematic and hermeneutic literature reviews. A systematic literature review can be defined as “a form of secondary study that uses a well-defined methodology to iden-tify, analyze and interpret all available evidence related to a specific research question in a way that is unbiased and (to a degree) repeatable” (Kitchenham 2007, p. vi). A hermeneutic literature review instead focuses on the understanding by the reader and deems the review as a creative, iterative and ongoing technique based on interpretation and the previous knowledge of the reader. The aim of the hermeneutic approach is to uncover all relevant knowledge independent of the source (Boell and Cezec-Kecmanovic 2011). In this research, we conduct a systematic and descriptive literature review as proposed by Rowe (2014) supplemented by elements of the hermeneutic ap-proach according to Boell and Cezec-Kecmanovic (2014) to answer our first research question.

In the work at hand, we apply Rowe’s approach as the leading research structure because it offers a practicable guideline in the form of seven steps that incorporate all aspects


51

that we consider relevant for our research. For the first step, Rowe recommends select-ing appropriate research questions for further investigation followed by selecting sources, choosing search terms, applying practical screening criteria, applying method-ological screening criteria, doing the review and synthesizing the results (2014). The research questions of our literature review are specified in the first section of this paper. The remaining steps are carried out in the subsequent sections. Boell and Cezec-Kecmanovic (2011) criticize the linear character of systematic literature reviews and the predefined inclusion and exclusion of sources and search terms. They argue that biases may arise if potential sources for literature are predefined based on prestige of journals or language and relevant literature is, therefore, excluded ex-ante. The hermeneutic ap-proach assumes that ultimate understanding can never be achieved, but a saturation in understanding can be reached by passing through the process. Boell and Cezec-Kecmanovic (2014, p. 264) structure the literature review as two intertwined circles, the wider analysis & interpretation circle and the narrower search & acquisition circle. The hermeneutic and the systematic approaches are not mutually exclusive, but in contrast they are supposed to complement each other. The hermeneutic approach helps to focus on understanding, broadening the search, and interpretation of results. Thus, we use it complementarily to identify all relevant papers. The reason for combining the ap-proaches into an iterative procedure is that the hermeneutic thoughts provide valuable components for the refinement of the search in multiple iterations. An additional ad-vantage of the combination is that a systematic review is particularly suitable where a lot of literature exists, while the hermeneutic approach broadly explores topics when it is assumed that no excessive body of knowledge exists. Although CA in general is well settled, its application to BI is very specific and a recent topic. This situation motivates a combination of both approaches to better approximate comprehensiveness of the re-viewed publications during the modified iterations of our linear search.

A.4 Literature Review This section describes the iterations of our review. Each iteration introduces the search procedure, presents, and analyzes the results. The procedure is an instantiation of the steps two to five proposed by Rowe (2014). After each reading phase, a mapping or classifying is followed by a critical assessment. If new vital elements are identified, the search is refined (Boell and Cecez-Kecmanovic 2014).


A.4.1 First Iteration

Selecting Sources. According to Rowe, preferably only top outlets should be considered “because lower rank journals do not address theory” (2014, p. 247). However, Rowe also states that if the topic is rather technical, conference contributions should be in-cluded. In respect of the specific interdisciplinary topic, we decided not only to search the basket of eight (Association for Information Systems 2011), but include all IS and Finance & Accounting journals according to Harzing’s Journal Quality List (2015) as well as top management journals according to Barreto (2010). The Journal Quality List was chosen for the identification of journals since it consolidates established interna-tional journal rankings. In addition, we cover the top IS conferences in our search: the European Conference on Information Systems and the International Conference on In-formation Systems.

Choosing Search Terms. In the conceptual foundations, we identify relevant keywords to be used as search terms. The subsequent combination of terms for CA methods and BI was used as search string for the first iteration: ("cost allocation" OR "costing" OR "internal allocation" OR "billing" OR "pricing" OR "assessment" OR "activity based costing" OR "relative direct cost" OR "internal activity allocation" OR "charging" OR “chargeout” OR "chargeback" OR "overhead rates") AND ("Business Intelligence" OR "Data Warehouse" OR "data warehousing" OR "analytics").

Applying Practical and Methodological Screening Criteria. Based on the selection of our sources and search terms, we only searched publications in English language. We did not restrict the search period in order to not exclude relevant results ex-ante. The search was performed for abstracts, titles and full texts. Further, we conducted forward and backward search (Webster and Watson 2002). The inclusion and exclusion of results was conducted as follows: results returned from the search were checked for plausibility of titles first. After passing the plausibility check, abstracts were reviewed for thematic orientation, and subsequently the full texts were analyzed when the abstracts were con-sidered relevant.

Table 9 shows the number of hits yielded from the search in journals and confer-ences.Table 10 summarizes results uncovered in the first iteration which are considered to be relevant for our research.


53

Table 9. Results of First Search Iteration

Search type Total hits Relevant hits Journal search 3,506 (4,790 ) 9 Conference search 364 (371 ) 3 Sum 3,870 12

Table 10. Analysis of First Search Iteration

# Reference Analysis 1. (Choudhary and Vithayathil

2013) Transfers concept of cloud computing to BI. Reveals that profit center organization is best for cloud services and BI.

2. (Hosanagar et al. 2005) Pricing strategies for caching services. Recommend different prices for different service qualities.

3. (Arnott and Pervan 2008) Justification of research topic. 4. (Watson et al. 2004) Management of BI should be done by a BI committee contra-

dicts the idea of allocations. 5. (Verner et al. 1996) Differentiation of CA methods for IS. 6. (Clark Jr et al. 2007) Justification of research topic. 7. (Schieder and Gluchowski

2011) Justification of research topic.

8. (Even and Shankaranarayanan 2006)

Influence of cost-utility considerations on the design of a DWH. No contribution on CA.

9. (Rosenkranz and Holten 2007)

Paper contains insights gained from a transition project from a pricing model to ABC for information technology (IT) ser-vices in a bank.

10. (Tallon et al. 2013) Chargeback mechanisms are an important part of IT govern-ance and promote accountability.

11. (Granlund 2011) Justification of research topic. 12. (Lönnqvist and Pirttimäki

2006) Introduce measurement approaches for BI value determina-tion and BI process management. Costs-value considerations.

None of the found publications covers the topic of BI CAs in particular. While some papers justify research in the field of IS cost management (e.g., Granlund 2011), others apply allocation methods to other domains, (e.g., IS in general or caching services). Some of the findings reveal insights on the antecedents of BI CAs. The first paper (Choudhary and Vithayathil 2013) puts emphasis on the organizational form of coordi-nation and recommends a profit center organization for the BI department. The second paper (Hosanagar et al. 2005) provides us with the idea that prices for BI services could vary according to the quality of the offered services. The differentiation of CA methods

Before removal of duplicates


according to different cost types in paper nine (Rosenkranz and Holten 2007) is a valu-able insight that may be considered for BI. Several publications justify our research topic of BI CAs. In summary, the results of the search in top rated scholarly journals are not satisfying in terms of comprehensiveness for the BI domain.

A.4.2 Second Iteration

The publications found in the first iteration treat the topic either from a generalized IT management perspective or on a domain level. Therefore, in the second iteration we explored the topic from a more abstract level to discover whether more general findings on CA for IS can be instantiated to BI. In the second iteration, our goal was to obtain an overview of literature on CAs for enterprise-wide IS (Laudon and Laudon 2006) with the objective of analyzing probable effects and interdependencies that can be derived from the higher level of generalization. We use the same outlet selection as in our first iteration. In this iteration, we combined the search terms for CAs (cf., first iteration) with: information system*, IT service* and IT cost* and search in abstracts and titles. Table 11 shows the number of total hits and relevant hits. The relevant results are sum-marized in Table 12.

Table 11. Results of Second Search Iteration

Search type Total hits Relevant hits Journal search 102 (220*) 14 Conference search 13 (15*) 1 Sum 115 15

Table 12. Analysis of Second Search Iteration

# Reference Analysis 1. (Tang and Cheng 2005) Derivation of different optimal prices depending on the

cost composition of web services. 2. (Westland 1992) Description of influences of external effects on the pricing

for IS and the related impacts on IS management. 3. (Ross et al. 1999) Ten cases on applied CA methods, impacts on IS manage-

ment, conditions and implications. 4. (Belcher and Watson 1993) Conduction of a value assessment and categorization of

costs. Emphasize importance of cost-benefit analysis. 5. (Zou and Huang 2013) Focus on importance and impact of service level agree-

ments in a pricing model for IT services. 6. (McKinnon and Kallman

1987) Mapping of the CA system to the organizational environ-ment depending on various factors.


55

7. (Gerlach et al. 2002) Case procedure for determining costs of IT services in or-der to implement ABC.

8. (VanLengen and Morgan 1993)

Findings indicate a relationship between a sophisticated CA method and measures of IS maturity.

9. (O’Connor and Martinsons 2006)

Present drivers of CA design.

10. (Olson and Ives 1982) Influence of CA system on user involvement. 11., 12.

(Peppard et al. 2007), (Karimi et al. 2001)

Highlighting the importance of CAs for the management of IS.

13., 14. 15.

(Brignall 1997), (Reid and Smith 2000), (Chenhall 2003)

Relationship between CA design and contingency theory.

The first (Tang and Cheng 2005) and the fifth publication (Zou and Huang 2013) illus-trate that a CA may differentiate prices and use service level agreements or catalogues. The research of Belcher and Watson (1993) adds the value perspective to the topic of CAs and categorizes the costs according to certain cost types. Ross et al. (1999) give detailed insights in ten cases that show that CAs are dependent on use situation and goals. In the findings, the goals of a CA are a recurring factor with influence on the design of a CA and its management impacts (e.g., Olson and Ives 1982; Verner et al. 1996). Paper six (McKinnon and Kallman 1987) presents the compelling thought of designing an CA depending on different situational factors, (e.g., system maturity). Con-sequently, the application of a CA method seems to be dependent on situational contexts (Brignall 1997; Chenhall 2003; Gerlach et al. 2002; O’Connor and Martinsons 2006; Reid and Smith 2000) and system maturity levels (VanLengen and Morgan 1993). Sum-marizing, the topic is not dealt with in a sufficiently generalized way allowing for adop-tion and adaption entirely to BI, but the findings provide expedient conceptions for fur-ther consideration.

A.4.3 Third Iteration

Referring to the hermeneutic circle, in our previous reading we identified additional journals specialized on the topics of BI or cost management. In this iteration, we searched in the journals: International Journal of Business Intelligence Research, Busi-ness Intelligence Journal, International Journal of Business Intelligence and Data Min-ing, International Journal of Data Warehousing and Mining, Journal of Intelligence Studies in Business, Ontology-supported Business Intelligence, Cost Effectiveness and Resource Allocation, Cost Management, Journal of Performance Management, Journal


of Cost Management. We searched on the EBSCO database, used the search string de-fined in the first iteration, and applied it to titles and abstracts. The search led to 88 hits. Unfortunately, none is considered relevant. All hits discuss how CA can be supported by BI, not how they can be applied to BI.

A literature review is a potentially never-ending circle because in every iteration, new knowledge is identified or interpreted upon which a new iteration can be designed (Boell and Cecez-Kecmanovic 2014). One option to finalize the search is saturation, (i.e., the fact that new iterations do not create additional relevant insights). The missing additional results may be interpreted as such saturation.

A.5 Literature Synthesis and Future Research

A.5.1 Literature synthesis

While the importance of CAs in IS management is shown (e.g., Tallon et al. 2013; Verner et al. 1996), some publications particularly call for research on BI cost manage-ment (e.g., Arnott and Pervan 2008; Schieder and Gluchowski 2011). No scholarly pub-lication focuses on BI CAs, but several instances that apply CAs to contiguous domains provide notions for a potential BI CA. Thus, no publication answers RQ1 completely, but we obtained interesting aspects for further consideration:

• All identified CA methods are applied to sub-domains of IS (e.g., Hosanagar et al. 2005; Tang and Cheng 2005; Watson et al. 2004), e.g., caching services or DWH. Nevertheless, one common characteristic is the situational application, e.g., depending on the IS maturity (VanLengen and Morgan 1993), organiza-tional environments (McKinnon and Kallman 1987), and external (Westland 1992) as well as internal contingencies (Brignall 1997; Reid and Smith 2000), which leads us to the assumption that there is no “one size fits all” approach to a BI CA.

• Verner et al. (1996) describe eleven effects of CA on IS management, e.g., re-source regulation or performance evaluation. Scattered works refer to the effects of IS CAs on IS management in general (Ross et al. 1999; Westland 1992). Other publications highlight the importance of CA methods for IS management (Karimi et al. 2001; Peppard et al. 2007). Olson and Ives (1982) stress the influence of IS CA on user involvement. Thus, desired management impacts from a CA – in


57

other words: the management objectives – call for particular consideration in a BI CA.

• A number of papers deal with antecedents of CAs for IS. While Ross et al. (1999) found that, in ten cases, IT costs were allocated based on total costs, Belcher and Watson (1993) contribute a cost categorization, and Gerlach et al. (2002) show a cost determination for ABC. The differentiation of CA methods according to dif-ferent cost types (Rosenkranz and Holten 2007) is a valuable insight. Two publi-cations contribute the ideas of price differentiation (Tang and Cheng 2005) and the use of service level agreements or catalogues (Zou and Huang 2013). Regard-ing the form of coordination, Choudhary and Vithayathil (2013) state that a profit center is more appropriate than a cost center organization. Another antecedent is the measurement of BI success and cost-utility considerations, for which hints are found in literature (Even and Shankaranarayanan 2006; Lönnqvist and Pirttimäki 2006; Schieder and Gluchowski 2011). O’Connor and Martinsons (2006) present several drivers for the design of a CA. The antecedents found are configurable components that shall be considered in a BI CA.

While the topic of CA is considered very relevant for BI systems, only few publications contribute specific aspects to the topic and no work could be identified that comprehen-sively addresses BI systems. There may be different reasons for this situation. First, researchers’ access to empirical data may be limited because cost data is highly sensi-tive, but relevant and rigorous research needs to incorporate practical experience (Hevner and Chatterjee 2010; Stokes 1997). Second, the establishment of BI systems can only be dated back to recent years, while the long-tail of research with interdiscipli-nary views on BI is still to come. The above synthesized results of existing work deliver a comprehensive foundation for future research on CAs for BI, but they also show that still many gaps need to be filled.

A.5.2 Fields for future research

The lack of literature that addresses CA in the domain of BI implies that either the ap-plicability of existing CA methods has not yet been researched under consideration of the specific characteristics of BI systems, or that specific methods need to be developed or adapted for this domain. The literature review illustrates that for a BI CA no “one size fits all” approach seems to be applicable. Therefore, future research shall consider a


situational contexts in order to integrate different use situations. In conclusion, we derive the following fields of future research (RQ2) according to three propositions:

1. Proposition: the foundations of BI CAs form the common ground that applies to all BI CAs in all different situational contexts and are the first step of further re-search:

The fact that no substantial contribution particularly dedicated to CAs for BI can be found leads us to the assumption that future research shall address the characteristics that make BI special for CAs.

Design principles for a BI CA shall provide a first basic artefact for a purposeful BI CA. Principles of form (Gregor et al. 2013) shall address the general structure, principles of function (Gregor et al. 2013) shall represent the functioning, and principles of imple-mentation (Gregor and Jones 2007) shall guide how to implement a BI CA in specific contexts.

2. Proposition: the situational context defines different use situations in which BI CAs are applied since the design of CAs are highly dependent on the goals and contextual factors:

The implementation of a CA is not done as an end in itself, but for a certain purpose, (i.e., to contribute to management goals). Therefore, the goal-impact relation shall be subject to future research to identify the goals of a BI CA and how the desired impacts can be realized. An empirical investigation on the goals and impacts of CA applied to BI shall lead to the profound understanding of the impacts on BI (and indirectly IS or even corporate) management.

The use situations of a BI CA are characterized by different contextual factors, (e.g., the BI system maturity, type of BI service). Thus, one major field of future research is to identify and evaluate the different situational contexts that need to be incorporated in the design of a BI CA. An empirical study of different existing BI CAs in practice shall evaluate applied methods based on in-depth case studies, derive a generic method by means of inductive reasoning and classify the approaches by describing their situational differentiations.


59

3. Proposition: the interaction between certain antecedents – representing individ-ually configurable components – specifies the configuration of BI CAs:

The literature synthesis shows that research on selective antecedents on CA exists, but research on interrelations of those antecedents that play a significant role for CA for BI is missing. In particular, the components derived from the literature review are the or-ganizational form of coordination, the cost types to allocate, and the types of BI services. A derivation of analogies from CAs in other domains (e.g., information pricing in finan-cial markets) can contribute valuable inferences to BI.

Since, along with the management of BI costs, several sources claim for measurement of BI success and cost-utility measures, value considerations in cost accounting for BI need to be researched. An essential part of measuring BI success is the derivation of meaningful key performance indicators. This is a particularly interesting since BI cost occurrence and BI value contribution differ regarding their granularity, association with an organizational unit and distribution over time.

The integration of the above mentioned fields of future research shall lead to a solution approach that assists practitioners with a decision-support for designing an appropriate BI CA. One prerequisite for this sketch of a solution approach is the ability to configure the BI CA according to the specific requirements of the applying organization.

A.6 Conclusion This paper represents the first extensive overview of the body of knowledge on CA methods for BI and works from related fields of research. For this purpose we present relevant CA methods existing in accounting literature and delineate the field of BI cost allocations. Further, we conduct a systematic literature review in three iterations en-riched by elements of the hermeneutic approach and analyze a large number of search results in our iterations. In the synthesis of relevant results we argue that our research question one is not yet sufficiently covered by existing literature, and in consequence from the obtained insights we derive areas of future research. Especially, we propose substantial research on a situational BI cost allocation method that incorporates founda-tions, the different use situations, contextual factors, and the configuration components.

However, we acknowledge several limitations that apply to our research. First, the re-striction to certain search terms may exclude relevant results containing differing nam-ing conventions and languages, but to the best of our knowledge we derive relevant


terms in the conceptual foundations and broadly address the field with our applied search terms. Second, potentially transferrable approaches from other application domains might also have been excluded.

Part B: Management Objectives and Design Principles for the Cost Allocation of Business Intelligence

61

Paper B – Management Objectives and Design Princi-ples for the Cost Allocation of Business Intelligence

Table 13: Bibliographical Information for Paper B

Title Management Objectives and Design Principles for the Cost Allocation of Business Intelligence

Authors & Affiliation Johannes Epple, Stefan Bischoff, Stephan Aier

University of St. Gallen, Institute of Information Management, Mueller-Friedberg-Strasse 8, 9000 St. Gallen, Switzerland [email protected], [email protected], [email protected]

Publication Outlet Pacific Asia Conference on Information Systems (PACIS) 2015

Publication Type Conference Paper – Completed Research



Rating (VHB2 Jourqual 3) C

Abstract

Today business intelligence (BI) systems, which provide management with decision-supportive information, are considered to be a prerequisite for organizational success. In contrast to the operation of BI, BI system management is still an emerging topic in information systems (IS) research. Even though the cost management of BI systems is highly relevant for practice, the field is widely unexplored. Cost allocations for BI sys-tems are supposed to enhance transparency, create cost awareness and support the man-agement of resources of the BI system. In our research we have conducted two focus group studies to examine the basis for BI cost allocations. First, we derive management goals and design principles for a BI cost allocation from an exploratory focus group. In a second step, we evaluate the goals and the design principles in a confirmatory focus group. Our research provides valuable insights on the application of BI cost allocations from our focus groups and contributes a basis for the design of BI cost allocation meth-ods.


62 Part B: Management Objectives and Design Principles for the Cost Allocation of Business Intelligence

Keywords

Cost Allocation, Business Intelligence, Focus Group, BI Management.

B.1 Introduction Today BI is acknowledged as an umbrella term for “technologies, applications and pro-cesses for gathering, storing, accessing and analyzing data to help its users make better decisions” (Wixom and Watson 2010, p. 14). BI systems are supposed to create value through the analytical use of the information obtained from these systems. BI is thereby considered to be a prerequisite for organizational success (Wixom and Watson 2010). Beyond the supposed value of BI, the expenses for BI are still growing by significant rates, e.g. 8% in 2013 (Gartner Inc. 2014b). However, the measurement of BI is consid-ered to be a difficult task, but it is crucial for justifying BI investment decision as well as for managing BI processes (Lönnqvist and Pirttiäki 2006).

The cost of BI are in most cases overhead costs, i.e., costs that are not directly attribut-able to a product or service sold. In any case, there is a high need to internally allocate the costs of BI further, either to the cost object that is sold to an external customer or to the internal organizational unit that caused the costs. Cost accounting is a sub-discipline of management accounting and comprises among others the topic of allocating costs to costs objects or to those organizational units that caused the costs (Rao 2007). Cost al-location is supposed to fulfil various purposes e.g., to enhance transparency for the growing amount of BI costs, to provide the correct basis for calculation, to create cost consciousness, to uncover inefficiencies in the use of resources, or to create desired con-trol effects for the management of the resources whose costs are allocated to other units (Klesse 2007). Therefore, cost allocation for BI costs is an appropriate means to enhance transparency and contain desired control effects for the management of the growing portion of BI costs.

The topic of BI cost allocations is located at the interface between cost accounting and IS research, and moreover, BI services (Vargo and Lusch 2008) appear to be of partic-ular interest to cost accounting and IS research, due to a BI system’s nature. First, the output of a BI system is information for which traditional cost-oriented internal pricing mechanisms for goods and services do not apply. Inherent to this characteristic is that the usefulness and the benefits of a BI system depends on the decisions made based on the obtained information rather than in the use of the system itself (Benbasat and Zmud 2003). Second, the intention of a BI cost allocation method cannot be the restriction of


63

use of resources, but needs to encourage users to use the system to the best of its capa-bilities. Thus, BI cost allocation should contribute to promoting the use of the BI system and not distract users from the system use. Further, the aspect of not distracting BI users is also important in the light of the voluntariness of use of a BI system. Third, the oper-ation of BI systems involve a large monolithic cost block (Bischoff et al. 2014) that cannot be managed without transparency about the causes of the costs.

The review of related work (cf., section B.2) reveals that only scattered attempts to build BI cost allocation methods exist and no comprehensive work could be identified, alt-hough several sources call for future research in the field of BI cost management (cf., section B.2). Therefore, our research aims at answering the following research questions (RQ):

1. What are the management objectives for BI cost allocation? 2. What are the principles for designing a BI cost allocation method that addresses

the underlying management objectives?

The answer to RQ1 is of particular interest, because the definition of the objectives of BI cost allocation is the essential starting point for the design of a BI cost allocation method. The design principles (RQ2) provide guidance for instantiating specific BI cost allocation approaches for given situations. We derive our findings based on empirical data of an exploratory focus group (EFG) conducted with BI specialists from five major banks. The findings of the EFG are evaluated in a confirmative focus group (CFG) with a different group of BI managers and BI specialists testing the findings’ validity. We add to the knowledge base on BI cost allocation and BI management and inform BI managers systematically designing BI cost allocation instances.

This paper is organized as follows: section two presents related work and provides the conceptual foundation. In section three we illustrate the research method applied in this paper. In section four we present the results of the EFG and in section five we provide the evaluation based on the CFG. In section six we discuss our findings, their implica-tions and the limitations of our research.

B.2 Related Work We have conducted a systematic literature review according to Rowe (2014) as well as a hermeneutic literature review according to Boell & Cecez-Kecmanovic (2014). The need for future research on BI cost management is discussed in various publications,


e.g., in order to foster executive and user commitment to BI (Clark Jr et al. 2007). Schieder and Gluchowski (2011) constitute that future research efforts are justified “to bring us closer to a long sought after means to assess and compare cost and benefit aspects of BI solutions” (p. 12). Our review of prior work uncovered that in practice-oriented journals scattered publications on the application of certain BI cost allocation approaches can be found (e.g., Grytz 2014). In addition, consulting companies publish papers stating the maturity of the topic of IT cost allocation (Deloitte 2011) and BI cost allocation in particular (Steria Mummert 2013) is still low. Consequently, we derive that there is a relevance of future research from the scientific community as well as from practice.

The basis for BI cost allocation can be found in accounting literature. In cost accounting research cost allocation aims at transferring costs internally from the unit providing the goods or services to the consuming units (e.g., from BI provider to BI consumer). In accounting literature the following general cost allocation methods are differentiated: no cost allocation (Verner et al. 1996), overhead rates (Coenenberg et al. 2009; Verner et al. 1996), internal activity allocation (Coenenberg et al. 2009; Verner et al. 1996), activity based costing (ABC) (Kaplan and Cooper 1998), and other more distinct meth-ods like relative single cost calculation (Ewert and Wagenhofer 2011). These allocation approaches are discussed under a variety of synonymously used terms and they are ar-chetypes that need further adaptation in order to be applicable to BI systems.

Although cost accounting has been widely researched in the management science of the 20th century and many theoretical groundwork has been published (e.g., Clark 1923; e.g., Cooper and Kaplan 1988; Kaplan 1984; Shillinglaw 1989; Vatter 1950), as a matter of course today’s current phenomena of IS research have not been considered in that era. From the perspective of IS research cost allocation gained recognition in the context of IT performance management. Hence, research on cost allocation of IT services in general (e.g., Laudon and Laudon 2006; McKinnon and Kallman 1987; Rom and Rohde 2007; Ross et al. 1999; VanLengen and Morgan 1993; Verner et al. 1996) and within various IS domains (e.g., Brandl et al. 2007; Hosanagar et al. 2005; Müller et al. 2011; Tang and Cheng 2005; Watson et al. 2004) can be found in literature. One common characteristic observed in the various use cases is that the applied cost allocation ap-proach heavily depend on the situational context and its influencing factors, e.g., the system maturity, form of organization, or the type of services to be allocated. Thus, there is no “one size fits all” approach neither to IS cost allocation in general, nor to BI cost


65

allocation in particular. Therefore, a BI cost allocation approach has to be discretely designed for the specific situation.

A BI cost allocation serves specific purposes and is no end in itself. The rationale behind a BI cost allocation is to achieve desired management impact in order to accomplish certain management objectives in specific situations. Existing literature partially pro-vides conceptions about the underlying objectives of a BI cost allocation. Klesse (2007) describes the goals of a cost allocation for data warehouse (DWH) services, Ross et al. (Ross et al. 1999) describe conditions and implications of IT cost allocations, and Müller et al. (2011) build a matrix of appropriate allocation approaches and goals of lean man-agement, but no research on the objectives of BI cost allocation can be found in literature so far.

The review of prior work revealed that only few sources give specific recommendations or guidelines on what to consider in the design of a cost allocation method for BI or IS, respectively. Verner et al. (1996) provide an overview of the different cost allocation methods for IS. Choudhary and Vithayathil (2013) state that the recommended organi-zational form of coordination for cloud computing is a profit centre and transfer the concept to BI. Various sources give examples for the implementation of a specific cost allocation method in a given situational context (e.g., Grytz 2014; Rosenkranz and Holten 2007; Watson et al. 2004). Klesse (2007) describes basic requirements and pre-requisites for the design of a cost allocation for DWH services and IT services, e.g., cost effectiveness, exactness and practicability of the cost allocation method. However, due to the special nature regarding costs, value, and use perspectives of analytical IS and BI (Chen et al. 2012) in contrast to other domains of IS the existing approaches are not applicable to BI in full extent. In summary, several sources contribute aspects to con-sider when designing a BI cost allocation method, but a comprehensive set of design principles for BI cost allocation is missing in literature.

B.3 Research Method The paper at hand follows the design science research (Hevner et al. 2004; Winter and Baskerville 2010) paradigm aiming at providing a solution approach to the above de-scribed real-world design problem. While RQ1 is purely answered by assessment and description from the EFG and subsequent evaluation in the CFG, it aims at delivering a solid understanding of characteristics for the design of a BI cost allocation method by


differentiating management objectives for a BI cost allocation. The answer to RQ2 com-prises the design principles that are derived by interpretation from the observations in the EFG. The result of RQ2, the design principles, is a solution to a real-world design problem which is referred to as an artefact (March and Smith 1995) in design research. Subsequently, we introduce the method for our exploratory and CFGs and the artefact type of design principles.

B.3.1 Focus Group

We follow a focus group-based approach to collect qualitative data. Qualitative tech-niques are appropriate for the development of theory (Bryman 1999). The strength of qualitative methods is the “capacity to explore human subject motivation and actions within a research study frame of reference” (Debreceny et al. 2002). Further, the quali-tative method of a FG supports the purpose of our research, because biases of individuals can be mitigated and consensus in the FG can be measured (Morgan 1997). Further, a FG offers the opportunity to discuss opinions and attitudes of the participants in an in-teractive setting. Consequently, the FG is particularly suitable for our purpose, because we aim at deriving new knowledge for an existing design problem in a controlled envi-ronment.

In our research we apply FGs to derive and evaluate management objectives and design principles for BI cost allocations. By means of an EFGs the researchers can “achieve rapid incremental improvements in artifact design” and “demonstrate the utility of the design” in a CFG (Tremblay et al. 2010, p. 602). According to Sonnenberg and vom Brocke (2012) FGs are an appropriate method in every evaluation step of the design process.

According to Tremblay et al. (2010) the participants of the EFG as well as the CFG shall have similar characteristics and be familiar with the field to which the solution artefact is applied – in our case BI. Tremblay et al. (2010) propose an eight-step approach for conducting a FG comprising the steps (1) formulate research problem, (2) identify sam-ple frame, (3) identify moderator, (4) develop and pre-test a questioning route, (5) re-cruit participants, (6) conduct focus group, (7) analyse and interpret data and (8) report results. In our research we follow the approach proposed by Tremblay (2010) for the conduction of our FGs.


67

B.3.2 Design Principles

In design science research different artefact types are proposed by March and Smith (1995): constructs, models, methods, and instantiations. Beyond these four artefact types, Gregor and Hevner (2013) suggest additional design science research artefact types: design theories, design principles, and technological rules.

In IS literature design principles are established as an artefact type and sub-types of design principles are differentiated. Markus et al. (2002) define two sub-types of prin-ciples: “principles governing the development or selection of system features and prin-ciples guiding the development process” (p. 185). In our research we follow the defini-tion provided by Gregor and Jones (2007) and Gregor et al. (2013) who distinguish:

• principles of form: refer to the general structure (e.g., the shape or architecture) of the design solution (Gregor et al. 2013)

• principles of function: refer to the general functioning (i.e., what it does and how it functions) of the design solution (Gregor et al. 2013)

• principles of implementation: refer to the process of implementation of the design solution in a specific context in practice (Gregor and Jones 2007)

Generally speaking, design principles shall give advice for the effective design and im-plementation of a solution in a specific context.

B.4 Exploratory Focus Group In this section we first describe the general setup and the procedure of the EFG. Subse-quently, we discuss the assessment of the participants’ objectives for a BI cost allocation (RQ1) and derive the design principles (RQ2) from the EFG.

B.4.1 Focus Group Setup and Procedure

As described in section B.3.1 we follow Tremblay et al.’s (2010) eight-step approach of FG design. Subsequently, we briefly present the setup and the procedure according to the eight steps.

(1) Formulate research problem: the formulation of the research problem was pre-pared by the researcher in advance to the EFG as presented in sections one and two of the paper at hand. The goals of the EFG were to collect empirical data regarding RQ1 and to derive design principles from the discussions (RQ2).


(2) Identify sample frame: regarding this step three decisions were taken. The re-search should comprise one EFG for artefact design. For the EFG a higher num-ber of participants was desired than for the subsequent CFG in order to obtain rich insights from a broad sample. For the EFG a sample size of approximately 20 persons was desired. As type of participants we defined BI specialists as well as middle to high BI management. Thus, we could obtain the data from people who are directly involved and who are responsible for BI cost allocation.

(3) Identify moderator: it was decided that one of the researchers takes the role of the moderator for the session. Following the recommendation of Tremblay et al. (2010) and Miles & Huberman (1994) and for reasons of the sample size two co-researchers were employed as further observers taking minutes of the discussion.

(4) Develop and pre-test a questioning route: the agenda and the questioning route for the meeting was set up by the researchers in advance and discussed with other faculty members. The questioning route contained 13 questions for each partici-pating company. The questioning route contained direct and indirect questions that aimed at obtaining general information about the BI cost allocation methods as well as at identifying goals of a BI cost allocation and general design patterns. Due to the high complexity of the topic the questions were distributed in advance to the participants to enable preparation to a certain extent. In addition, the par-ticipants of each of the five major banks were asked to prepare an introductory presentation about their BI cost allocation approach and the advantages, disad-vantages and challenges they are currently facing. At this step the procedure dif-fers from the pure form of the eight-step approach (Tremblay et al. 2010) which does not include the possibility of preliminary conversations in preparation of the FG (Krueger and Casey 2000).

(5) Recruit participants: for the recruitment of participants we used contacts from an ongoing project, in the course of which workshops with BI representatives from major banks are conducted at regular intervals. The BI representatives were in-vited to kindly participate in our FG research project. Table 14 provides an over-view of the participants of our EFG. The group consisted of very experienced managing BI staff with a deep understanding of the processes in their organiza-tion. In Table 15 the company profiles of the participants’ banks are summarized. Due to their size each of the five banks has a significant BI organization or even several distributed BI units, respectively.


69

(6) Conduct focus group: the FG took place in June 2014 during a regular meeting of the workshop community. The FG lasted for three hours and was held in Ger-man language. The fact that the participants already knew each other from previ-ous sessions of the workshop made the get-to-know part of the FG obsolete and an atmosphere comfortable for an open discussion was guaranteed. For the ses-sion a seating at tables in U-shape was established which should be fun to the participant and stimulate active participation (Stahl et al. 2011). For the opening of the session an impetus presentation was held by the moderator which already fostered an active discussion. Subsequently, one or two representatives of each bank introduced their BI cost allocation approach and the challenges they are facing. Afterwards, the answers to the preliminary questions were introduced by each bank and subject to further discussion. Two researchers independently took notes, which were consolidated with the conceptions of the moderator after the FG. The documentation of the contributions by several researchers and the sub-sequent consolidation avoids misinterpretations (Miles and Huberman 1994), serves for documentation purposes and ensures reliability. After the EFG a sum-mary of the discussions was sent out to the participants asking for addition com-ments or objections.

Table 14: Participants of EFG

Participants of exploratory focus group # Company Current position Years of experience in BI 1 Bank A Deputy BI competence centre manager 22 2 Bank A BI solution manager 9 3 Bank A Assistant to BI competence centre manager 4 4 Bank A Deputy BI competence centre manager 17 5 Bank A BI self-service manager 6 6 Bank A Assistant to BI competence centre manager 8 7 Bank A Technical data warehouse project manager 27 8 Bank A BI solution manager 29 9 Bank A BI solution manager 13 10 Bank A BI solution manager 23 11 Bank B Head of Enterprise Architecture Management 15 12 Bank B BI solution manager 26 13 Bank B BI solution manager 28 14 Bank C DWH/BI architect 32 15 Bank C Head of BI architecture 10 16 Bank D Head of BI competence centre 13 17 Bank E BI development manager 8 18 Bank E IT organization – Head of BI 14


Table 15: Company Profiles of Participants of EFG

Company profiles – participants of EFG # Bank Employees 2013 Balance sheet total 2013 1 Bank A approx. 60,000 approx. $ 1,000 bln 2 Bank B approx. 50,000 approx. $ 600 bln 3 Bank C approx. 48,000 approx. $ 900 bln 4 Bank D approx. 44,000 approx. $ 250 bln 5 Bank E approx. 30,000 approx. $ 500 bln

The steps (7) analyse and interpret data as well as (8) report results are subsequently conducted and presented in the section B.4.2 for RQ1 and in section B.4.3 for RQ2.

B.4.2 Management Objectives of BI Cost Allocations

In the following, we first present the management objectives of BI cost allocation pre-vailing in the companies of our EFG. Afterwards, we discuss the results obtained.

Bank A has a long established BI landscape with a vast amount of BI systems and a very broad range of offered BI services. The applied BI cost allocation methods have grown with the BI landscape and have been continuously adapted. To date, Bank A runs a va-riety of different cost allocation methods for different BI services comprising internal activity allocations based on a service catalogue as well as different service level agree-ment offerings. Further, several assessments with different actual and planned consump-tions (overhead rates) are in place. The objectives for conducting BI cost allocations for Bank A are to allocate the costs according to the cost-by-cause principle, enhance transparency and to control the efficient use of resources. In addition, for Bank A BI cost allocations are supposed to contribute to exploiting BI to its full potential by drawing conclusions about usage behaviors.

Bank B employs a decentralized BI landscape with several BI applications running in the business units. The central BI department provides the backend architecture and sup-ports the business units as a business partner. The only BI cost allocation method em-ployed is an assessment based on planned consumptions. The objectives of the BI cost allocation for Bank B are to enhance transparency and to create cost awareness among the BI consuming business units.

The BI landscape of Bank C is on a relatively high maturity level with well-established BI systems across the entire bank. In contrast to Bank A, Bank C only uses the cost allocation method of an assessment based on actual consumptions to allocate BI costs.


71

They further differentiate different BI applications for which the costs are allocated by the assessment cycles. Bank C’s objectives for employing the BI cost allocation are to enhance transparency, to allocate the costs according to the cost-by-cause principle and to earn a profit for the BI department which should be used for future invest-ments.

The BI competence center of Bank D is very much driven by external pressure from the regulatory environment. Most of the current BI applications are installed only to comply with regulatory requirements, e.g., BCBS 239, Basel III or AnaCredit. Nevertheless, the BI costs are allocated to the BI using departments – in the case of Bank D the units that have to fulfill the regulatory requirements. The used allocation method is an assessment based on planned consumptions. The pursued objectives of Bank D are to enhance transparency, to create cost awareness and to allocate the costs according to the cost-by-cause principle.

Similar to Bank B, Bank E also has a much decentralized BI landscape in place. Bank E uses the most rudimentary BI cost allocation method with an assessment based on planned consumptions that do not reflect the actuals consumptions by far. The main objective of Bank E for the BI cost allocation is to credit 100% of the BI department and to debit the BI using business units. Further, Bank E aims at creating cost trans-parency. However, the allocated costs play no role for the debited departments, because they are shown “below the line” and have no influence on the target structure and incen-tive system.

Although the accomplishment of the objectives lies beyond the scope of the paper at hand, it needs to be mentioned that in the majority of the cases the employed BI alloca-tion methods are not appropriately designed to achieve the objectives set. The partici-pants of the EFG unanimously agreed that in every bank there is a high potential for improvement regarding the design of the BI cost allocation method in order to achieve the management objectives.

All of the participants approved that creating cost transparency is one of the main objectives of their BI cost allocations, since the BI services are very complex by nature and cause high costs which are often subject to internal discussions in the companies. Participant #14 from Bank C mentioned that for Bank C “the creation of cost transpar-ency for BI costs is the major objective, because especially for BI the management does not know for what exactly the costs occur and how much they spend in particular for the


different BI services.” The further objectives of creating cost awareness in the BI con-suming departments, allocating the BI costs according to the cost-by-cause principle and exploiting BI to its full potential are complementary with the creation of cost trans-parency.

Although the BI departments of all of the banks are organized as cost centers, the par-ticipants from three banks confirmed that for the BI department it would be helpful to earn profits for future investments. The remaining participants argued that this is not the business purpose of the BI departments, especially since they are all treated as cost centers, and in a profit center organization this objective might be appropriate. Partici-pant #13 stated that this wish is an “attempt to avoid budgeting discussions” which are normal to take place in every company, but which are especially hard to justify for BI, because of the intransparent character. The participants agreed that the through a cost allocation created transparency could also assist for the budgeting discussions, but all participant would prefer to be able to earn profits instead of long-lasting budgeting dis-cussions.

The discussions about the objectives of Bank E showed that, if the main objective of a BI cost allocation is 100% crediting of the BI department, it is mutually exclusive with the other objectives, because it leads to a simplified allocation method that cannot achieve further objectives. In the case of Bank E the BI cost allocation method is em-ployed as an end in itself.

The results from the EFG regarding RQ1 show that the objectives for a BI cost allocation are partially congruent with the objectives found in our review of prior work (e.g., Klesse 2007; Müller et al. 2011), but become a special context for BI cost allocations. Further, the objectives of earning profits and 100% crediting of the BI departments are particularly applicable in specific design situations. Therefore, we conclude that an evaluation in the CFG shall further enlighten our results.

B.4.3 Design Principles for BI Cost Allocations

From the discussions in the EFG about objectives and specific designs of BI cost allo-cation combined with our review of related work we derive design principles of BI cost allocation. We distinguish principles of form (cf., Table 16) and principles of function (cf., Table 17). Subsequently, we present the principles and provide complementary elu-cidations.


73

Table 16: Principles of Form for a BI Cost Allocation Method

# Principles of form 1 The BI cost allocation method must be transparent and traceable for all parties in-

volved. 2 The allocated costs for BI must be visible and manageable for the receivers. 3 Overly complex allocation logics need to be avoided as they are counterproductive and

not economic. 4 A profit center is the preferred form of organization, but the form of organization of the

BI unit has to be aligned with the overall management accounting and steering logic of the company.

5 If an internal activity allocation is applied to allocate BI costs further, a service cata-logue and service level agreements should be in place.

6 The selected allocation method should incorporate the BI system’s contextual factors.

Principle of form 1: EFG revealed that the BI cost allocation method shall enhance the transparency about the BI costs. This principle is targeted at a higher level of abstraction, i.e., the cost allocation itself. From the discussions in the EFG we conclude that in many cases the allocation logic is not clear to the people involved in the allocation on sender-, receiver- and executing-side.

Principle of form 2: the participants of the EFG argued that if a BI cost allocation is in place, it must trigger a certain steering/ management impact for the receivers. Therefore, the visibility of the allocated costs for the receivers is a further important factor.

Principle of form 3: among the participants the common sense evolved that the cost allocation method shall be economic. Although BI is a complex domain, too complex allocation methods are considered to be counterproductive.

Principle of form 4: while the participants argued that the under prevailing conditions they have not the right to earn profits, it was agreed that they would prefer to be orga-nized as a profit centre. Participant #4 argued that “if we were organized as a profit center, the BI consumers would directly be aware how much the BI services cost and think about the value of the additional information or their requests. Thus, cost-value considerations for BI would be automatically established.”

Principle of form 5: this principle is a result of the discussion about the allocation logic of Bank A whose allocation logic is currently very advanced. The participants of the remaining banks agreed that the practice of the cost allocation logic of internal activity allocation can be considered best practice.


Principle of form 6: from the practice of Bank A it also became evident that the prior consideration of contextual factors, like e.g., system maturity or management objectives, leads to superior results in the design of the BI cost allocation method.

Table 17: Principles of Function for a BI Cost Allocation Method

# Principles of function 7 The BI cost allocation method should not restrict users from system use, and should

not negatively influence value generation through BI, but foster use of the BI system and services.

8 A BI cost allocation should not be an end in itself, but be a means to realize certain management goals.

9 The transparency about BI costs and BI services must be clearly enhanced through the cost allocation method.

10 The cost allocation of actual costs should not be used to fund future BI projects in a cost centre organization. Profit centres and investment centres can “sell” their BI ser-vices with margin.

11 The allocation logic should also be applied to the planned values. The BI provider and the BI consumer must be held responsible for the variances.

Principle of function 7: this principle is very crucial for the allocation of BI costs. Participant #16 stated that “the usage of BI must not be restricted by the cost allocation. BI is different from other internal services, since the value through BI is created by the decisions made based on the obtained information.“ He gave the compelling example that “for internal catering services it makes sense to restrict the usage, but not for BI.”

Principle of function 8: the participants agreed that the objective of 100% crediting of the BI department is not sufficient for designing a BI cost allocation method, but it must have an impact.

Principle of function 9: it was unanimously decided by the participants that every BI cost allocation should enhance the transparency about BI costs and BI services.

Principle of function 10: this principle is the group’s common denominator after the discussion about earning profits in a BI department the participants.

Principle of function 11: participant #5 asked the group what their measures for per-formance evaluation on sender and receiver side are. After discussing how the partici-pants calculate their plan values and enriched by a best practice example of Bank A the group agreed that the allocation logic of the actuals also needs to be applied for planned values and the variances are the primary performance measures.


75

The presented design principles seem not in all cases to be BI-specific, but in some instances generally applicable to cost allocations. However, these are the results of the EFG, which need further evaluation in the CFG.

B.5 Confirmatory Focus Group In this section we again first present the setup and the procedure of the CFG. Subse-quently, we discuss the results of the evaluations regarding RQ1 in section B.5.2 and regarding RQ2 in section B.5.3.

B.5.1 Focus Group Setup and Procedure

For the CFG we also follow the eight-step approach according to Tremblay et al. (2010). Below, we introduce the specifications of the steps for the CFG.

(1) Formulate research problem: the research problem for the CFG was to evaluate the management objectives and the design principles derived from the EFG in order to confirm and/or refine the results and designs from the EFG (Hevner 2007).

(2) Identify sample frame: for the CFG we aimed at recruiting a highly specialized participants in order to obtain a focused discussion. Therefore, the sample should consist of experienced personnel from BI management and the size should not exceed the size of the EFG. In contrast to the composition of the EFG, it was desired to recruit participants from diversified industries to eliminate the bank-ing-specific bias.

(3) Identify moderator: as moderator for the CFG the same researcher as for the EFG was chosen. In the CFG again two further researchers were installed as observers taking notes. One of the two observes was a co-researcher, the other one was an independent researcher to ensure objectivity.

(4) Develop and pre-test a questioning route: the agenda and the evaluation route for the CFG was set up and reconciled with the co-researchers and the independent researcher prior to the CFG. First, an impetus presentation was held in order to establish a common understanding for the very specialized topic of BI cost allo-cations. To avoid bias in the subsequent discussion of the participants the topics of management objectives and design principles were directly and indirectly ex-cluded from the presentation. The questioning route regarding RQ1 first con-tained an assessment of the management objectives for a BI cost allocation of the


participants. Afterwards, the management objectives from the literature review and from the EFG were presented to the CFG and a gap analysis with a discussion was conducted. In regards to RQ2 the single design principles derived from the EFG were discussed individually with the participants of the CFG and evaluated in the dimensions comprehensibleness, practicability, importance, and BI speci-ficity. The choice of the dimensions aimed at evaluating the importance of the principles, if it is generally applicable or BI-specific, and if the participants con-sider it to be successfully realized. At the end of the session the participants had to complete a questionnaire with their demographic information.

(5) Recruit participants: as the platform for conducting the CFG the researchers had the possibility to use a highly specialized practitioner-oriented conference on BI and DWH. The session was announced in the conference program to be for the target audience consisting of BI leaders and management accountants. The par-ticipants had to subscribe to the session in advance to enable the researcher to verify the suitability of the group for conducting the CFG. In total we obtained 13 subscriptions in advance, whereof eleven participants attended the session.Ta-ble 18 gives an overview of the participants and their company profiles. The in-dustry backgrounds of the participants are very heterogeneous, but all partici-pants had a relevant management role in companies with significant BI organi-zations.

(6) Conduct focus group: the CFG was conducted in November 2014 during the above mentioned conference for BI practitioners. It lasted 90 minutes and was held in German, although the material for presentation and evaluation was pre-pared in English. Due to the relatively small group and the impetus presentation the moderator accomplished to establish a cozy working atmosphere already at the beginning of the session. In the interactive part for the evaluation of manage-ment objectives and design principles a very lively and open discussion took place, granting rich insights in the practices and angles of the participants. The co-researcher and the independent researcher took notes of the discussion and consolidated them with the conception of the moderator after the session. Some of the participants left their contact details to obtain a wrap-up of the session and to be engaged in more detailed analysis.


77

Table 18: Participants and Company Profiles of CFG

Participants of confirmatory focus group

# Current position Years of expe-rience in BI Industry Total sales of par-

ticipant’s company 1 Head of BI competence centre 15 Insurance € 10-15 bln

2 Head of BI solutions 11 In-house consult-ing

> € 100 bln

3 BI manager responsible for management accounting solu-tions

12 Logistics € 2-5 bln

4 Head of BI competence centre 18 Government € 101-500 mln

5 Deputy head of BI competence centre 17 Insurance € 101-500 mln

6 BI Senior manager 9 Consulting € 10-15 bln 7 Senior BI product manager 15 IT vendor € 50-100 bln 8 Head of BI competence centre 14 Banking € 20-50 bln 9 Partner/ head of BI solution 11 Consulting € 20-50 bln 10 BI manager 8 Automotive > € 100 bln 11 BI project manager 10 Utility € 50-100 bln

The steps (7) analyse and interpret data as well as (8) report results are conducted and presented in the subsequent sections.

B.5.2 Evaluation of Management Objectives

As already described first the management objectives for a BI cost allocation of the participants of the CFG were assessed. Subsequently, the objectives obtained from the EFG as well as from an extensive literature review were discussed. In Table 19 an over-view of the management objectives is provided. Surprisingly, the objectives (1)-(5) were directly confirmed by the CFG, whereof the participants unanimously attached the great-est importance to the enhancement of cost transparency, due to the same reasons like the participants of the EFG. Participant #1 put special emphasis on the importance of en-hancing cost transparency about the cost for BI caused in decentralised departments. Further, the objective of exploiting BI’s full potential of BI was unanimously confirmed and attached with high importance. However, participants #3 and #6 argued that exploit-ing BI’s full potential is kind of wishful thinking, and the objective should be renamed to better exploiting the potential of BI.


Table 19: Evaluation of Management Objectives

Evaluation of management objectives for BI cost allocations

Objective Assessed in EFG

Confirmed in CFG Existing basis in literature

(1) Enhance cost transparency Yes Yes Yes (e.g., Müller et al. 2011) (2) Create cost awareness Yes Yes Yes (e.g., Müller et al. 2011) (3) Cost-by-cause allocation Yes Yes Yes (e.g., Müller et al. 2011) (4) Exploit BI’s full potential Yes Yes - (5) Efficient use of resources Yes Yes Yes (e.g., Müller et al. 2011) (6) Earn profit for BI dept. Yes Partially - (7) 100% crediting of BI dept. Yes No - (8) Restrict use of resources No Yes Yes (e.g., Müller et al. 2011)

The objective (6) was only partially confirmed by the CFG, because the participant dis-cussed that the insights from a BI cost allocation shall be used for issues regarding fi-nancing and investment decisions. At this point the participants reached the same com-mon denominator like the participants of the EFG, but without arguing about the right for BI departments to earn profits. The objective of 100% crediting of the BI department was not confirmed by the CFG. After presenting the objective to the group, the consen-sus of the group was clearly that this should not be an objective for designing a BI cost allocation method, but in some cases the BI cost allocation gets “degenerated” (partici-pant #3) to that objective. In addition, the objective of restriction of use of resources was brought up by participant #1, who argued that in their company the use of the central BI function is sometimes misused for unnecessary activities, e.g., design of unnecessary reports. This newly added objective is similar to the objective (5), but clearly aims at cutting the use of resources down to the necessary level. Since it only aims at restricting users from nonessential activities it is not contradictory to design principle #7.

B.5.3 Evaluation of Design Principles

The evaluation of the design principles was conducted in an interactive session in two groups. Therefore, we printed the design principles and the evaluation dimensions (com-prehensibleness, practicability, importance, and BI specificity) on posters and pinned them on two flipcharts. The two groups discussed the design principles independently with one of the researchers.

The consolidated results show that all of the design principles got an average to high ranking in terms of comprehensibleness. The outcome regarding the verbalization of the


79

principle is very satisfactory for us, since some of the understanding problems were due to translation difficulties and could be erased right away during the session.

The evaluation dimensions of practicability and importance are exclusion criteria, i.e., a low-rated evaluation in this dimensions leads to the elimination of the design principle. On the basis of low ratings and the interpretation of the discussions the design principles #4 (due to non-practicability) and #11 (due to unimportance/ implicitness) are elimi-nated. The remaining design principles all withstand the evaluation in regard to their importance and practicability.

The fourth evaluation dimension aims at distinguishing generalizable and BI-specific design principles. According to the evaluation the design principles #1, #3 and #8 are considered to be generally valid to IS cost allocations. The participants discussed that those principles are applicable to IS costs, and are probably transferable to other areas, but definitely not to all departments, e.g., for production departments different principles apply.

Consequently, the following (cf., Table 20) six design principles are considered to be especially important and applicable to a BI cost allocation.

Table 20: Design Principles for a BI Cost Allocation Method

# Principles of form 2 The allocated costs for BI must be visible and manageable for the receivers.

5 If an internal activity allocation is applied to allocate BI costs further, a service cata-logue and service level agreements should be in place.

6 The selected allocation method should incorporate the BI system’s contextual factors.

# Principles of function

7 The BI cost allocation method should not restrict users from system use, and should not negatively influence value generation through BI, but foster use of the BI system and services.

9 The transparency about BI costs and BI services must be clearly enhanced through the cost allocation method.

10 The cost allocation of actual costs should not be used to fund future BI projects in a cost centre organization. Profit centres and investment centres can “sell” their BI ser-vices with margin.

B.6 Discussion and Conclusion We base our research on an extensive review of existing literature and present relevant prior work. The applied research method follows the paradigm of design science, Fur-ther, we introduce the theoretical foundation for conduction of our FGs as well as the


artefact type of design principles. From the EFG conducted with 18 experienced partic-ipants in the domain of BI in five major banks we derive six management objectives and eleven design principles. In a CFG with eleven participants of senior BI management from diversified industries the management objectives and the design principles are evaluated. The conduction of the FGs is accurately planned, performed in a lively at-mosphere and rigorously analysed. In conclusion, we present a compelling list of eight management objectives for a BI cost allocation, six BI-specific design principles for a BI cost allocation method, and three general design principles for (IS) cost allocations.

Our research closes the identified gap existing in research on BI cost management. We contribute a solid basis for the design of a BI cost allocation method containing an ex-ploration and evaluation of objectives and design principles for a purposeful design of a BI cost allocation method. The presented results deliver compelling answers regarding the existing real-world design problems regarding the objectives of a BI cost allocation, and provide assistance to BI managers and management accountants by presenting sci-entifically validated design principles. Further, certain generally valid design principles provide guidance for the design and application of cost allocations even in other (IS) domains than BI. The presented findings can be applied in practice, e.g., by taking the design principles into consideration as a “checklist” during the conceptual design of a (BI) cost allocation method.

The presented research is subject to a number of limitations. First, as inherent to research comprising qualitative data, different kinds of biases might slip in. Due to the fact that the EFG only consisted of employees of banks, there might be an industry bias in the data obtained from EFG. Further, subjectivity and misinterpretations are a potential risk adherent to qualitative data. However, to the best of the authors’ knowledge the industry bias is excluded from the interpretation of the results. The installation of several re-searchers taking notes independently (Miles and Huberman 1994) was definitely a valid countermeasure against potential misinterpretations and biases. Second, the study can-not claim for comprehensiveness, because it “only” comprises two focus group studies. Therefore, further research needs to revisit the topic. The decision to focus on particular parts of the BI cost allocation is due to the fact that almost no prior work contributing to the advancement of the topic can be found. Further, the high complexity of the topic requires a breakdown into sizable sets.

Nevertheless, future research shall address several aspects of a BI cost allocation that remain, due to the limited scope of this research project. First, research on contextual


81

factors further influencing the design of a BI cost allocation method, e.g., BI system maturity or status of BI system acceptance, shall draw a comprehensive picture of the different design situations. Second, research on the method fragments, e.g., cost types, organizational aspects, and allocation mechanism shall provide further guidance for the purposeful design and configuration of a BI cost allocation method.

Part B: Contextual Factors Influencing the Purposeful Allocation of Business Intelligence Costs

83

Paper C – Contextual Factors Influencing the Purpose-ful Allocation of Business Intelligence Costs

Table 21: Bibliographical Information for Paper C

Title Contextual Factors Influencing the Purposeful Allocation of Business Intelligence Costs

Authors & Affiliation Johannes Epple

University of St. Gallen, Institute of Information Management, Mueller-Friedberg-Strasse 8, 9000 St. Gallen, Switzerland [email protected]

Publication Outlet European Conference on Information Systems (ECIS) 2016




Rating (VHB3 Jourqual 3) B

Abstract

Today, business intelligence (BI) systems are recognized as a prerequisite for organiza-tional success by providing management with decision-supportive information. Along with the pervasion of BI in organizations and its prospective benefits, BI has become widely established from a technological perspective, but remains challenging from an organizational view. Because BI causes a high monolithic cost block within an organi-zation, effective BI cost management mechanisms need to be implemented. This study focusses on the different design situations of BI cost allocations since a “one size fits all” approach is not supposed to be purposeful, but different situational requirements need to be considered. The design situations are characterized by relevant contextual factors, which we identify and thoroughly analyze. A set of relevant contextual factors is derived from an extensive review of prior work as well as from a focus group study with subject-matter experts. The contextual factors are synthesized in a classification scheme according to the dimensions objectives, technology-, organization-, and envi-ronment-related. Further, we discuss the characteristics of contextual factors and the impacts on cost allocation configurations in three archetypes of companies. Therefore,


84 Part B: Contextual Factors Influencing the Purposeful Allocation of Business Intelligence Costs

this publication delivers valuable insights for consideration in practice and stimulates further research on context-specific BI cost management.

Keywords

Business-IT Alignment, Cost Management, Contextual Factors, BI Management.

C.1 Introduction By supporting and improving the decision-making process in organizations, business intelligence (BI) systems are widely considered to be a prerequisite for organizational success (Wixom and Watson 2010). Even though, BI has a considerable impact on the decision capabilities of an organization’s management (Audzeyeva and Hudson 2015) by turning the vast amount of data into decision-supportive information, the manage-ment of BI costs and value is especially difficult. On one hand, the determination of BI’s return on investment is challenging (Negash 2004), since inherent to BI is a downstream value because the output of BI is data converted into information that leads to (probably) beneficial business decisions. On the other hand, BI causes high upfront costs in its in-vestment phase (Bischoff et al. 2015) as well as a large monolithic cost block in its operations (Bischoff et al. 2014) that cannot be managed without transparency about the causes of the costs. While general information technology (IT) expenses only rose by 0.4% in 2013 (Gartner Inc. 2014a), BI expenses grew significantly faster in 2013 with 8% (Gartner Inc. 2014b). Therefore, purposeful cost management mechanisms are es-sential to realize value from BI investments and effectively control BI costs.

Information systems (IS) cost management is an interdisciplinary topic between IS gov-ernance and cost accounting, which is a sub-discipline of management accounting (Rao 2007). Within the field of IS cost management, cost allocations (CA) that charge costs from the IS providing unit back to the IS consuming business unit are a core activity (Van Grembergen and De Haes 2009) that offers a variety of prospective benefits. Ef-fective CAs can serve multiple management purposes, e.g., providing information for product costing, performance evaluations, informing investment decisions, creating transparency or uncovering inefficiencies in the use of resources (Kaplan and Cooper 1998; Shim and Siegel 2000). Cost management has been widely researched in the man-agement science of the 20th century (e.g., Cooper and Kaplan 1988; Shillinglaw 1989; Vatter 1950) and a reception of CAs has taken place in IS research. A prominent exam-ple emphasizing the role of IS CAs is the widely known COBIT (Control Objectives for


85

Information and Related Technologies) framework (Information Systems and Control 2005) that explicitly describes a process to identify and allocate costs.

Although several publications highlight the importance of IS CAs (e.g., Karimi et al. 2001; Peppard et al. 2007) for IS management, still many facets of IS CAs remain un-explored. According to a survey by Deloitte (2011), the maturity of IS CA is still low in practice. Stefanov et al. (2012, p. 1) attest that IS CA is “still poorly understood” and a lack of successful allocation models can be observed. Several contributions on IS CAs focus on the application of CAs in dedicated domains of IS, e.g., caching services (Hosanagar et al. 2005) or web services (Tang and Cheng 2005). One common charac-teristic is that there is no “one size fits all” approach to an effective IS CA; rather a high number of differently configured CAs depending on the application-specific domain and on the organizational situation in which the CA is applied. Due to the fact that the situ-ation can be characterized by different contextual factors, it is crucial to uncover and classify the contextual factors relevant for the configuration of a BI CA. Consequently, this paper addresses the following research question (RQ):

What are the contextual factors influencing the purposeful configuration of a BI cost allocation?

Answering the RQ closes the research gap regarding the different design situations for BI CA. Our findings are based on extensive review of prior work and extended by qual-itative data from a focus group (FG). In a practical sense, our findings contribute to a better understanding of different design situations, which assists practitioners making BI CAs successful by recognizing implications of contextual factors. The remainder of this paper is structured as follows: section two introduces the conceptual foundations of our research to establish a common ground and to define the research gap. In section three, we illustrate the overall research process, in which the paper is embedded, and provide the research methods applied here. According to the introduced research meth-ods, section four provides the results from our literature review, and section five adds the results of our FG study. In section six, we synthesize the results from the sections four and five, present a classification scheme of the contextual factors, depict exemplary combinations of contextual factors, and discuss the implications on BI CAs. We discuss our findings, their limitations, and implications on future research in section seven.


C.2 Conceptual Foundation

C.2.1 Cost Allocations

Basic CA mechanisms are introduced in Table 22.

Table 22: Basic Cost Allocation Mechanisms

Mechanism [Synonyms] Description (Source) No cost allocation The costs are not further allocated, but remain as overheads

with the provider (Olson and Ives 1982) Overhead rates [assessment] A key (e.g., number of users) is used as a distribution rate

to allocate the costs for BI to the consumers (Verner et al. 1996)

Internal activity allocation [billing, chargeout, chargeback]

Prices for products or services are defined. The consumers are debited with used quantity * price (Verner et al. 1996)

Activity based costing Process costs and cost drivers for the processes are calcu-lated. Upon use, the consumers are debited with the costs (Kaplan and Cooper 1998)

Relative direct cost calculation Being a rather special type for CAs, it attributes cost to cost objects that are related to the decisions causing the costs (Ewert and Wagenhofer 2011)

A CA charges costs for the internal consumption of goods or services from one organi-zational unit (e.g., a cost center) to another organizational unit or to a cost object (e.g., a production order) by crediting costs to the provider and debiting the consumer with a certain amount of costs. A CA mechanism refers to the underlying mechanism to trans-fer costs from provider(s) to the consumer(s). The differentiation of CA mechanisms is crucial at this point because it needs to be considered in the configuration of BI CAs.

C.2.2 Business Intelligence

The concept of BI is understood in a broad sense because it encompasses all components of an integrated decision support infrastructure (Baars and Kemper 2008). This includes the technical and the social components, or as Herschel (2010, p. i) puts it “today, the practice of BI clearly employs technology. However, it is prudent to remember that BI is also about organizational decision-making, analytics, information and knowledge management, decision flows and processes, and human interaction.” The comprehen-sion of BI in the paper at hand includes the technical aspects with analytical frontend applications, a data warehouse (DWH) and the interfaces to operational data sources. Further, it encompasses the organizational process view including the sociotechnical


87

perspective. Thus, not only the costs of the technical components of BI (e.g., hardware and software) are subject to a BI CA, but also the costs incurring in the execution of the related processes and the sociotechnical interactions (e.g., labor costs and external ser-vices).

BI appears to be of special interest in IS CAs due to a BI system’s nature. First, the usefulness of BI depends on the decisions made based on the obtained information rather than in the use of the system itself (Benbasat and Zmud 2003), which inheres a potential downstream value. Therefore, BI is difficult to be priced for CAs since neither the cost of production nor the value of the information can be easily determined. Second, ac-cording to Ross et al. (1999, p. 232), the untapped potential of an effective CA is “to educate business units about IT while teaching the IT unit about the business.” This potential benefit of a CA is especially important for BI, due to the fact that on one hand BI systems need to be continuously meeting the ever changing information needs of the business, e.g., in fulfilling ad-hoc reporting requirements or frequent data model adap-tions. On the other hand, for a purposeful use of BI business units need to better under-stand the set-up (e.g., the data model) and the capabilities of BI systems in place. There-fore, an effective CA contributes to the mutual understanding between IT and business units to exploit BI’s potential to a better extent. Third, in contrast to CAs in other do-mains, the intention of a BI CA is not a restriction of use of resources, but it needs to encourage users to use the system to the best of its capabilities. Thus, a BI CA bears the potential to regulate the economic use of a BI. On one hand, a BI CA should promote the voluntary use of BI systems and not distract users from BI use. On the other hand, a BI CA should also prevent from “over-analyzing” by critically reflecting the question “what to analyze?”. Therefore, simply implementing – a for IS costs widespread (Ross et al. 1999) – usage-based CA is not purposeful for BI because it can cause a "death spiral" of decreasing utilization. Therefore, several publications can be found that par-ticularly call for future research on BI cost management issues (Arnott and Pervan 2008; Clark Jr et al. 2007; Schieder and Gluchowski 2011).

C.2.3 Situational Context

Van Grembergen and De Haes state that “[…] governance of IT is a very broad concept and that each organization requires a specific approach applicable to its individual con-text” (2009, p. 201). The origins of situational and context-specific perspectives in man-agement can be traced back to Fiedler’s (1964) work about the contingency theory in


the field of social psychology. Other scholars (e.g., Reid and Smith 2000) argue that even earlier publications on contingency considerations exist (e.g., Burns and Stalker 1961; Chandler 1962). In contrast to prior theories, contingency theory emphasizes that there is no “one best way” of organizing. Instead, the contingency theory argues that there are different effective ways, depending on the organizational context, and – in other words – addresses “the multivariate nature of organizations and attempts to inter-pret and understand how they operate under varying conditions” (Wallace et al. 1980, p. 370). Thus, varying conditions are defined by internal or external contingency varia-bles that are referred to as contextual factors (Weill and Olson 1989).

Contingency theory and research on contextual factors have become established in the IS discipline (Raber et al. 2013a) and the adoption to IS research can be found in a wide range of publications (e.g., Bucher et al. 2007; Schonberger 1980). Weill and Olson (1989) identify seven established contextual factors in IS research: (1) size, (2) environ-ment, (3) strategy, (4) structure, (5) technology, (6) task, and (7) individual characteris-tics. Contextual factors have an effect on different units of analysis in IS research such as e.g., organizations, processes, tasks. Therefore, in the following, we identify the con-textual factors particularly relevant for BI CAs. These factors are informed not only by IS research, but also by reference disciplines of the given topic, which are accounting and organizational studies.

C.2.4 Research Gap

In this section, we build the relations between the above-described conceptual founda-tions and reflect the need for research presented in the paper at hand. As described in the introduction, the measurement of BI costs and value is considered to be a difficult task, but it is crucial for justifying BI investment decision as well as for managing BI costs and resources (Lönnqvist and Pirttiäki 2006). Therefore, practitioners need guid-ance for improving BI regarding questions of BI cost management, realization of enter-prise capabilities, and business/ IT alignment. In cost management, BI CAs are supposed to be an appropriate instrument to trigger desired steering impacts for the management of BI costs and resources. Several basic CA mechanisms exist, but they need further adaptation to be applicable to BI systems. The adaptation of a CA mechanism to a par-ticular subject (in our case BI) is considered as the configuration of a CA, which includes a broad variety of different settings, e.g., defining prices or keys, identifying sender-


89

receiver relationships, or definition of BI services and cost elements to allocate. More-over, it is important to note that a company could also simultaneously implement differ-ent CAs mechanisms for different purposes, e.g., overhead rates to allocate BI infra-structure costs combined with an internal activity allocation for defined BI services. Therefore, the configuration of a BI CA is a complex task, which has not been appro-priately informed by theory yet.

In section C.2.2, it is shown that BI is a special case for CAs due a BI system’s nature, which needs to be considered in the configuration of a BI CA. In section C.2.3, the concept of situational context is introduced, arguing for the specific consideration of contextual factors in the configuration of a BI CA. Various sources give examples for the configuration of a specific BI or IS CA in a given situational context (e.g., Grytz 2014; Rosenkranz and Holten 2007; Watson et al. 2004), but a comprehensive work on the purposeful configuration of a BI CA, considering different design situations is miss-ing.

C.3 Research Method

C.3.1 Research Process

The results of the paper at hand are embedded in a research initiative with the overall research goal of:

designing a method for the configuration of BI CAs that incorporates the specific design situation in order to close the existing research gap and to support practitioners with the purposeful design of a BI CA.

The overall research method of our research initiative follows the design science re-search (DSR) paradigm, according to Peffers et al. (2007). DSR intends to solve existing real-world problems through the design of useful artifacts (Hevner et al. 2004; Winter and Baskerville 2010). Different artifact types need to be distinguished: constructs, mod-els, methods, and instantiations (March and Smith 1995; Winter 2008) as well as design theories, design principles, and technological rules (Gregor and Hevner 2013). For the design of the method we use the differentiation made by Bucher et al. (2007) in our research who characterize “design situations” and “method fragments” as the central components of the situational configuration. On one hand, the configuration of a BI CA is highly dependent on the situational context in which the allocation should be imple-mented. On the other hand, in every BI CA, various components have to be configured


(the method fragments) according to the situational context, e.g., the applied basic CA mechanism (cf., section C.2.1).

The paper at hand represents an essential step towards our proclaimed overall research goal because it contributes a classification scheme for different design situations. Ac-cording to Gregor and Hevner (2013), classifications are recognized as useful descrip-tive knowledge in DSR. Thus, in this paper, we aim at developing descriptive knowledge as part of an overall artifact – a method – that represents prescriptive knowledge. To develop the descriptive knowledge as an element of the overall artefact, we follow the steps one to four of Peffer’s proposed procedure (2007). To shed light on our RQ, a review of existing studies in the accumulated body of knowledge is an essential first step of our research. Therefore, we first conduct a literature review focusing on the iden-tification of existing publications on relevant contextual factors, which corresponds to Peffer’s first three steps “identify problem & motivate”, “define objectives of a solu-tion”, and “design and development” (2007). To extend the “design and development” step and to perform the “demonstration” step in order to demonstrate the usefulness of the results, we conduct an exploratory FG to collect real-world data from practice about contextual factors relevant for BI CAs. A FG is especially purposeful for our research because it offers the opportunity to explore further prevailing contextual factors that are considered relevant by BI experts affected by CA. FGs help to “achieve rapid incremen-tal improvements in artifact design”, to “demonstrate the utility of the design” (Tremblay et al. 2010, p. 602), and they offer the possibility to encourage and seize the interactions among the participants, while biases can be mitigated and consensus can be measured (Morgan 1997).

C.3.2 Literature Review

In the first step of our research, we conducted a comprehensive literature review com-bining the systematic literature review procedure according to Rowe (2014) with the hermeneutic literature review approach according to Boell and Cecez-Kecmanovic (2014). In this section, we briefly introduce the two approaches. In section C.4, we pre-sent the results identified in existing literature.

We applied Rowe’s seven-step approach (2014) as the guiding research structure be-cause it contains appropriate steps for our purpose: selecting appropriate research ques-tions for further investigation, selecting sources, choosing search terms, applying prac-tical screening criteria, applying methodological screening criteria, doing the review and


91

synthesizing the results. For further details, the readers are referred to the original con-tribution (Rowe 2014). In contrast, a hermeneutic review puts emphasis on the under-standing by the reader and considers the review as an iterative, creative, and ongoing technique based on interpretation by the researcher independent of the source because bias may occur due to the linear procedures of systematic approaches and the ex-ante defined search criteria (Boell and Cezec-Kecmanovic 2011). The hermeneutic approach is structured in two intertwined circles: the search & acquisition circle and the wider analysis & interpretation circle (Boell and Cezec-Kecmanovic 2011). The hermeneutic and the systematic approach are not mutually exclusive, but, on the contrary, the herme-neutic approach is supposed to complement and enrich the systematic steps by focusing on understanding, interpretation, and broadening the search.

We complementarily employed the two approaches in our iterative review in the follow-ing way: after each review step, a hermeneutic classification was followed by a critical assessment, which in turn led to a refinement (Boell and Cecez-Kecmanovic 2014) of search terms and sources applied in the following iteration. In the first and second iter-ation, we searched the basket of eight (Association for Information Systems 2011), all IS as well as Finance & Accounting journals according to Harzing’s Journal Quality List (2015), top management journals according to Barreto (2010), the European Conference on Information Systems, and the International Conference on Information Systems. In the third iteration, we included specialized BI and cost management journals. We only searched English publications. The search period was not restricted. The search was conducted for abstracts, titles and full texts. In addition, we applied forward and back-ward search (Webster and Watson 2002).

C.3.3 Focus Group

In our research, we followed Tremblay et al.’s (2010) eight-step approach of FG studies. Subsequently, we briefly present the setup of our FG.

Formulate research problem: in our FG we had the objective of exploring new contex-tual factors in addition to our literature review in order to collect empirical data regard-ing our RQ1 as well as discussing prevailing combinations of contextual factors and demonstrating implications on BI CAs.

Identify sample frame: according to Tremblay et al. (2010) the participants of the FG shall be familiar with the field in which the solution artifact is about to be designed. Thus, we defined BI specialists and BI managers, who are experienced with BI CAs, as


the desired participant types in order to obtain data from participants who are directly affected and who are responsible for the subject. The ideal FG consists of three to twelve participants and takes between one and three hours (Tremblay et al. 2010).

Identify moderator: one of the researchers was chosen to take the role of the moderator. Following Tremblay et al. (2010) and Miles and Huberman (1994), it was decided that two co-researchers are employed as observers, taking minutes of the discussions to en-sure objectivity.

Develop and pre-test a questioning route: prior to the FG the agenda as well as the questioning route were predefined by the researchers and reconciled with faculty mem-bers who are specialized on research in the field of BI management. The questioning route comprised questions about the general patterns of the applied BI CAs as well as direct and indirect questions to identify contextual factors.

Recruit participants: we decided to recruit the participants from the BI competence cen-ter of our faculty. In the BI competence center, regular benchmarking and knowledge sharing workshops with BI subject-matter experts from major banks of German-speak-ing countries take place. Table 23 provides an overview of the characteristics of the participants of the FG. The FG consisted of experienced managing BI staff with a thor-ough understanding of their organization and processes. The participants represented significant BI organizations or even various distributed BI units, respectively.

Table 23: Focus Group Participants

# Current position Com-pany

Employees 2013

Balance sheet total 2013

1 Head of Enterprise Architecture Manage-ment Bank A approx. 50,000 approx. $ 600 bn

2 BI department manager Bank A approx. 50,000 approx. $ 600 bn 3 Head of BI architecture Bank B approx. 48,000 approx. $ 900 bn 4 DWH/BI architect Bank B approx. 48,000 approx. $ 900 bn 5 Head of BI competence center Bank C approx. 44,000 approx. $ 250 bn 6 Project manager BI competence center Bank C approx. 44,000 approx. $ 250 bn 7 BI competence center, reporting manager Bank C approx. 44,000 approx. $ 250 bn 7 Head of BI development Bank C approx. 44,000 approx. $ 250 bn 8 BI development manager Bank D approx. 30,000 approx. $ 500 bn 9 IT organization – Head of BI Bank D approx. 30,000 approx. $ 500 bn

Conduct focus group: the FG study took place in December 2014, lasted for approxi-mately three hours, and was held in the German language, although the materials were


93

prepared in English. The fact that the participants are familiar with each other from pre-vious workshops made the first acquaintance part of the FG obsolete and immediately a comfortable atmosphere for open discussions was ensured. A circle-shaped seating was arranged with the intention of stimulating active participation (Stahl et al. 2011). At the beginning of the session, an impetus presentation on the research problem and the role of contextual factors for BI CAs was given by one of the co-researchers and the moder-ator to establish a common understanding. Subsequently, participants provided insights to the BI CAs in their banks and their organizational settings. A pleasant working at-mosphere fostered proactive and lively interactions, granting rich insights into the par-ticipants’ practices. After the initial discussions, the moderator led the discussions to RQ1. The minutes were consolidated with the moderator’s conceptions after the session in order to avoid misinterpretations, to serve for documentation purposes, and to ensure reliability of the results (Miles and Huberman 1994).

The steps (7) analyze and interpret data as well as (8) report results are presented in section C.5.

C.4 Results of Literature Review To synthesize the results of the literature review (Webster and Watson 2002) we relate them to our research according to the dimensions: management objectives and contex-tual factors related to technology, organization, and environment, which can be also found in prior works (e.g., Chenhall 2003; Reid and Smith 2000). Generally speaking, it became apparent that in recent years a resurgence of contingency considerations took place in accounting and IS research to cope with current technical phenomena, e.g., cloud services (Stefanov et al. 2012). Several contributions (e.g., Anderson and Young 1999; Weill and Olson 1989) provide overviews of prior work related to our research and many publications shed light on the topic from different angles (e.g., from an or-ganizational perspective or the influence on management control systems).

C.4.1 Management Objectives

Several publications (Drury 2000; e.g., Olson and Ives 1982; e.g., Ross et al. 1999) mention management objectives influencing the effective configuration of CAs. In ad-dition, we identified management objectives particularly applicable for BI CAs in a sep-arate contribution (Epple et al. 2015). We plead for a separate consideration apart from the other contextual factors, due to the special characteristic of objectives. In contrast to


other contextual factors, an objective has an intrinsically motivated origin rather than being internally or externally varying conditions that lie beyond the sphere of influence (Otley 1980). Management objectives inhere a common characteristic with other con-textual factors of being an object influencing the effective configuration of a CA. Table 24 subsumes the identified objectives according to the categories “use-/ resource-re-lated” and “cost-/ performance-related”.

Table 24: Overview of Management Objectives

Management objective (source)

Use

-/ re

sour

ce-

rela

ted

Efficient use of resources (Ross et al. 1999; Verner et al. 1996) Effective resource utilization (Drury 2000; Ross et al. 1999; Verner et al. 1996) Resource regulation (Epple et al. 2015; Verner et al. 1996) Exploit BI’s full potential (Epple et al. 2015) Increase user and IS staff awareness (Drury 2000; Verner et al. 1996) Increase user & management involvement (Verner et al. 1996)

Cos

t-/ p

erfo

r-m

ance

- rel

ated

Cost-by-cause allocation (Epple et al. 2015) Cost recovery (Olson and Ives 1982) Cost transparency (Epple et al. 2015; Verner et al. 1996) Performance evaluation (Drury 2000; e.g., Ross et al. 1999; Verner et al. 1996) Outsourcing evaluations (Verner et al. 1996) Better planning and budgeting (Verner et al. 1996)

It is crucial to apprehend in which way the single objectives influence an effective con-figuration, which we briefly discuss in the following. Except for the objective of cost recovery, all other objectives require a fairly detailed and accurate cost information in order to achieve the desired objective. In contrast, if only cost recovery is the single objective, a simple CA can be implemented (Verner et al. 1996). For the use-/and re-source-related objectives, which are dependent on the acceptance by the users, an inter-nal activity allocation based on activity prices with transparent pricing mechanisms might be appropriate. The objectives of BI department evaluation and outsourcing eval-uation cannot be successfully achieved without additional information, e.g., in terms of key performance indicators or benchmarks. In order to increase involvement or to raise awareness further governance mechanisms, e.g. accountability areas/levels, need to be in place. Further, a company might not strive to achieve only one of the above presented objectives, but rather pursue several concurrently. Certain objectives might even work mutually exclusive, while other objectives are complementary. Therefore, it is decisive to define and reflect the objectives in advance to the configuration of an appropriate CA.


95

C.4.2 Contextual Factors

Subsequently, we provide overviews of identified and relevant contextual factors from our literature review in three tables according to the three dimensions technology, or-ganization, and environment. Further, we briefly describe the impact of the contextual factors on the configuration of a CA. Following each table, we explain how the identi-fied factors apply for BI.

Technology is discussed as a contingency of CAs already since the early era of contin-gency theory (e.g., Bruns Jr. and Waterhouse 1975; Otley 1980; Wetherbe and Whitehead 1977) as well as in contemporary works (e.g., Abdel-Kader and Luther 2008; Auzair 2015; Chenhall 2003). Table 25 gives an overview of contextual factors related to technology.

Table 25: Contextual Factors Related to Technology

In BI research maturity considerations and maturity models are considered viable in-struments for the purposeful management of BI (Raber et al. 2013b). Thus, we conclude that the BI system maturity is a contextual factor relevant for the appropriate configu-ration of a BI CA. Since BI systems can in fact be employed for different purposes, e.g., supporting monthly closing activities, for turnaround information, or strategic decision-making, the type of BI system is further considered as a contextual factor. BI includes a “broad category of technologies, applications, and processes for gathering, storing, accessing, and analyzing data” (Wixom and Watson 2010, p. 14). Therefore, we con-

Contextual factor (source) Impact on the configuration of a CA

Tec

hnol

ogy

IS maturity (McKinnon and Kallman 1987; VanLengen and Morgan 1993)

Complexity of the CA rises with increased IS ma-turity.

Type of IS (McKinnon and Kallman 1987)

Four types of IS “support, factory, turnaround, strategic”, where a support IS requires the least so-phisticated CA.

Standardization and automation (Chenhall 2003)

The higher standardization and automation of the IS, the more formalized is the management control system.

Task uncertainties and task interde-pendences (Anderson and Young 1999; Chenhall 2003; Drury 2000)

High uncertainties and interdependences between technologies lead to more informal controls with less reliance on standard procedures and account-ing performance measures.

IS innovations (Reid and Smith 2000; Schonberger 1980)

Advancements of cost management measures are often preceded by significant innovations.


sider standardization and automation as well as task uncertainties and task inter-dependencies relevant contextual factor in our research. In contrast to other contextual factors, innovations are not a permanently influencing contextual factor, but rather spo-radically occurring. Due to the fact that BI is continuously further developed, we con-sider BI innovation a relevant contextual factor, which can influence the configuration of a BI CA.

In contingency research a big variety of works on organizational aspects prevail, due to the fact that contingency theory first was applied to questions of organizations research (cf., section C.2.3). Table 26 provides an overview of contextual factors related to or-ganization.

Table 26: Contextual Factors Related to Organization

A company’s strategy is often mentioned as a contextual factor related to organization (e.g., Chenhall 2003; Langfield-Smith 1997; Weill and Olson 1989), but we neglect it here, due to the fact that strategy traditionally is broken down and operationalized in specific goals (see section C.4.1). The contextual factor maturity of the IS function and IS management we consider as the maturity of BI management in our research context. Ross et al. (1999) and O’Connor and Martinsons (2006) mention several capabilities that influence the configuration of a CA, e.g., prevailing chargeback policies (consisting


Org

aniz

atio

n

Company size (Bruns Jr. and Waterhouse 1975; Chenhall 2003; Weill and Olson 1989)

With increasing company size the need for appro-priate cost accounting measures rises, due to in-creasing complexity, higher overall BI costs, and distributed accountability.

Maturity of the IS function and IS management (McKinnon and Kallman 1987; VanLengen and Morgan 1993)

The complexity of CAs rises with increased ma-turity of the IS function and IS management.

Maturity of cost accounting (O’Connor and Martinsons 2006; Ross et al. 1999)

Mature cost accounting organizations lead to higher accuracy, transparency, controllability, and fairness through a CA.

Degree of centralization (Hufnagel and Bimberg 1989; Olson and Chervany 1980; Raghunathan and Raghunathan 1992; Waterhouse and Tiessen 1978)

A higher decentralized IS structure requires more formal control mechanisms and administrative pro-cesses in IS budgeting, planning, and cost manage-ment. Thus, the need for effective cost manage-ment increases with decentralization.

(Top) management support (Anderson and Young 1999; McKinnon and Kallman 1987)

Proactive management support is more likely to re-quire sophisticated cost management measures and support complex CAs than a reactive management support.


97

of sourcing policy, cost recovery policy, policy on accountability) and administrative practices (rate setting process, communication process), which are related to the ma-turity of cost accounting. The degree of centralization of the BI function is regarded to be a relevant contextual factor for our research, since highly centralized as well as highly decentralized BI structures exist.

Table 27 provides an overview of contextual factors related to the environment of a company that need to be considered because external factors influence IS CAs as well as internal factors.

Table 27: Contextual Factors Related to Environment

In our research, we perceive legal or contractual obligation as a contextual factor di-rectly or indirectly influencing the configuration of a BI CA. If the obligations regulate the cost management information the company has to provide, a direct influence pre-vails. If the obligations prescribe certain BI structures, the CA is indirectly influenced. Further, we consider the combined contextual factor competition and uncertainty as well as the type of industry to influence the effective configuration of BI CAs because various publication show effects of these factors related to IS cost management.

It is essential to highlight that the ensemble and interplay of contextual factors needs to be considered in the configuration of a BI CA, since organizations are characterized by different contextual factors. Further, in particular the interplay of contextual factors re-lated to technology and related to organization appears to refer to the degree of business/ IT alignment. Therefore, the degree of business/ IT alignment can be regarded to as a subsuming set integrating certain contextual factors of the two categories. Consequently, a high degree alignment could allow for sophisticated BI CA mechanisms.


Env

iron

men

t

Legal or contractual obligation (Davis and Olson 1985; Verner et al. 1996)

Potentially leads to a configuration that fulfills the minimum requirements of the legal or contractual obligation.

Competition and uncertainty (Brignall 1997; Chenhall 2003; Gordon and Miller 1976; Khandwalla 1972; Otley 1980)

High competition can cause a focus on cost reduc-tions. An accurate and transparent cost manage-ment can contribute to a company’s ability to iden-tify cost cutting potentials.

Industry (Blanton et al. 1992) The type of industry (traditional or modern) and the related products or services influence IT and IS CA mechanisms.


C.5 Results of Focus Group In this section we analyze, interpret, and report the results of our FG study presented in section C.3.3. On one hand, contextual factors, which we already revealed in our litera-ture review, were confirmed in the FG. Due to space limitations we do not report details about the discussions on the confirmed factors. On the other hand, three new contextual factors were identified in the FG study. Subsequently, we discuss the newly identified contextual factors and present interesting aspects of confirmed factors.

The phase of acceptance of BI services was unveiled and debated by the participants as a factor influencing the configuration of BI CAs. In general, a consensus prevailed among the participants that the phase of acceptance shall be considered in a similar way like the maturity of BI services in the design of a CA. Participant #3 stated that “in lower phases of acceptance the use of BI services shall rather be subsidized to stimulate use and reach a higher acceptance.” In our analysis we agree, that a sophisticated CA might not be purposeful, and might even cause effects contrary to their intended results, if BI services are not even accepted yet. In the course of the discussion on phases of ac-ceptance, the participants brought the aspect of BI architecture into play, since some of the participants believed decentralized architecture to enjoy higher acceptance by users. In the participants’ companies different BI architectures exist, e.g., one enterprise DWH or several functional DWHs. Therefore, the participants recognized differences in their CAs resulting from differences in their BI architectures. Dissents remained on whether decentralized BI architectures enjoy higher acceptance. In our research, we take up the phases of acceptance for further consideration. In contrast, we do not explicitly take over BI architecture, since the discussions showed that the architectural aspect is inherent to the degree of centralization as the technological aspect of centralization.

In the discussion on centralization, the participants of bank D explained that within their company the BI department is monopolistic and does not compete with other internal or external providers. As a result, a basic BI CA is set up in bank D that “does not have appropriate steering impacts due to missing accountability for the results of the CA” (participant #9). Undisputedly, in this example the monopolistic status as a result of high centralization influences the configuration of BI CA.

According to the discussions, all participating companies are organized as cost centers. Nevertheless, the participants unanimously agreed that the corporate structure of cost accounting influences the configuration of BI CAs. According to participant #7 other


99

departments within their organization are organized as profit centers and employ more market oriented CA mechanisms. In addition, participant #5 stated that if BI is organized in a separate company, legal transfer pricing regulations influence the configuration of the CA. On one hand, the organization in cost centers, profit centers, investment centers, or separate companies is subject to the configuration of a CA. On the other hand, the BI organization is embedded in a given corporate cost accounting structure, which influ-ences the configuration of a BI CA. Therefore, we consider the corporate cost account-ing structure as a contextual factor.

Apart from bank C, which has a high CIO involvement, all participating companies re-ported about an issue between funding the original investment and bearing the ongoing operational costs. Even in bank C the funding is supposed to influence the later alloca-tion of operational costs. According to the participants the original funder becomes the owner of the BI application and has a stake in the allocation of operational costs. Thus, in case a BI application is centrally funded all users can be equally charged with opera-tional costs. In contrast, if a BI application is funded by one functional department, but used by several functional departments, the original funder must be able to partially recover its investment. Thus, we adopt the form of original funding as a contextual factor.

Participant #5 informed that in his perception in their company involved parties “have fun with finding the most appropriate allocation key”. A lively debate revealed different organizational cultures, e.g., a “free rider mentality” (participant #9) regarding BI costs in bank D. This discussion confirmed the contextual factor of maturity of cost account-ing, which is assumed to influence organizational cultures by setting appropriate cost accounting policies.

C.6 Synthesis of Results Synthesizing the results of our research we propose a classification scheme in form of a morphological box that consolidates all identified relevant contextual factors for BI CAs. A morphological analysis is an appropriate means for the synthesis of our results, since it is referred to as “a method for structuring and investigating the total set of rela-tionships contained in multi-dimensional, nonquantifiable, problem complexes” (Ritchey 2011, p. 7). Table 28 shows the morphological box including all contextual factors as well as example characteristics. In fact, we used the characteristics identified with the contextual factors in our literature review. If in the literature review revealed


no appropriate characteristics for a contextual factor, we derive the example character-istics from other research works related to the contextual factor. Regarding the maturity considerations exhaustive work in the area of BI exists by Raber et al. (2012), from which we adopt the denominations of the maturity levels because the focus of these work fits to our research interest. Regarding the characteristics of phases of acceptance we refer to Cooper and Zmud (1990), since it is a widely recognized conceptualization in IS research. For the newly identified contextual factors from our FG study we take over the discussed characteristics.

Table 28: Morphological Box Comprising Relevant Management Objectives and Con-textual Factors

In practice a big amount of different combinations of management objectives and char-acteristics of contextual factors might occur. Due to the fact that every assessment of a company according to the proposed classification scheme represents a snapshot at a cer-tain point and objectives and factors are not stable, even for a single company several design situations might be applicable over time. Further, probably one characteristic of a contextual factor indicates a certain configuration of a CA, but at the same time the characteristic of another contextual factor suggests a different configuration. Thus, it is important to demonstrate ideas how the insights on contextual factors can be applied, wherefore the vehicle of depicting archetypical companies is often used (e.g., Miller 1975; Reid and Smith 2000). Reid and Smith (2000) present three archetypical compa-nies: adaptive, running blind, and stagnant. An adaptive company dynamically operates,

Initiate Harmonize Integrate OptimizeSupport Factory

NoInitiation Adoption Adaption Acceptance Routinization Infusion

Initiate Harmonize Integrate OptimizeInitiate Harmonize Integrate Optimize

Corporate cost accounting structure Cost center Profit center Separate company OtherForm of original funding of BI system Central Other

Traditional, manufacturing

Traditional, manufacturing&service

Traditional, service

Modern, manufacturing

Modern, manufacturing&service

Modern, service

Investment centerSeveral fundersOne funder


1. Efficient use of resources 4. Exploit BI’s full potential 1. Cost-by-cause allocation 4. BI performance evaluation

Standardization and automation Low Medium High

2. Effective BI resource utilization 5. Increase user and IS staff awareness 2. Cost recovery of BI department 5. Outsourcing evaluation3. BI resource regulation 6. Increase user & management involvement 3. Enhance cost transparency 6. Planning assistance

PerpetuateTurnaround Strategic

Contextual factors related to technologyContextual factor Example characteristicsBI maturity levelType of BI system

Task uncertainties/interdependencies Low Medium HighBI innovations Seldom Frequent

(Top) management support Low Medium High

Phase of acceptance of BI serviceContextual factors related to organization

Contextual factor Example characteristicsCompany size Small Medium Large

PerpetuatePerpetuate

Maturity of BI managementMaturity of cost accounting Degree of centralization Centralized Decentralized

Contextual factors related to environmentContextual factor Example characteristicsLegal or contractual obligation Yes NoCompetition & uncertainty Less competitive Medium competitive Highly competitive

Industry


101

running blind company intuitively makes its decision, and a stagnant company is stable with conservative decision making. In the following, we draw three types of BI compa-nies based on real-world examples from our BI research and discuss implications of the contextual factors on the BI CAs. Example 1 refers to a company from the FG of the paper at hand. Example 2 arises from a practical BI project and example 3 has its origins in a prior FG study (Epple et al. 2015). These examples represent – in our understanding – archetypical combinations of contextual factors (characteristics are highlighted in ital-ics) with dominating influences of certain factors.

Example 1: the adaptive BI company. In example 1, we refer to a large sized global bank in a highly competitive market and legal obligations require certain BI structures. A broad BI landscape is fully centralized for data storage and processing and decentral-ized for reporting applications. Mainly mature and accepted BI services comprising dif-ferent types of BI systems exist. Maturities of BI, BI management, and cost accounting are considered on the level “optimize”. Frequent innovations are implemented. The bank is organized in cost centers and a differentiated funding model for BI components exists. Overall, high management support for BI prevails and an explicit commitment to all our identified objectives exists with emphasis on: effective and efficient use of resources, exploitation of BI’s potential to a high extent, cost-by-cause allocation, en-hance transparency, and performance evaluation. The CAs of example 1 appear highly sophisticated consisting of various combined CAs for different management purposes, e.g., internal activity allocations based on defined service catalogues and different ser-vice level agreements. Further, several overhead rates with different and accurate actual and planned keys are in place. In addition, a mark-up on full cost prices is charged on certain BI consulting services to earn reserve funds for future innovations driven by the BI department without dedicated sponsors. The BI CAs are subject to continuous revi-sion and adaption and contribute to use- and resource-related objectives, e.g., exploiting BI to its full potential, by drawing conclusions about usage behaviors. A high effective-ness of BI CAs is reached in regard to the predefined objectives. The driving contextual factors are supposed to be the mature technological and organizational aspects, the competitive environment as well as clearly defined objectives and management support. These factors also outweigh the lower requirements of legal obligations in the configu-ration of the BI CAs.

Example 2: the running blind BI company. This example refers to a large sized tra-ditional transportation company with a monopolistic market position. No contractual


obligations exist, but political pressure from public shareholders questioning high ex-penses – also in IT. A big BI department is partially centralized (for data storage) and decentralized for (partially redundant) data processing and reporting. Various types of BI systems with widely differing maturity levels exist. The different BI services in the expedient BI landscape enjoy different phases of acceptance. The corporate cost ac-counting is structured in cost centers and on the lowest maturity level. The management objectives are not clearly defined, but claim to target cost recovery and creation of cost awareness. Only a rudimentary CA is implemented using overhead rates (allocation key: computers per cost center), which is a means in itself and does not have any management impact. Thus, example 2 is running blind in terms of leveraging the potential of CAs. Due to increasing pressure from shareholders urgent advancements in cost management are necessary, which will affect BI cost management and are supposed to lead to a better exploitation of BI capabilities. In the running blind BI company a broad BI landscape with lacking integration and harmonization has grown. Missing cost management capa-bilities, low management support, and unclear objectives are dominant factors causing an ineffective BI CA.

Example 3: the stagnant BI company. This example refers to a medium sized innova-tive IT inhouse consulting in the automotive industry in a less competitive environment. The centrally provided and centrally funded BI services have medium automation and standardization, and are – apart from report adjustments – rarely changed. BI services are in the pre-acceptance phase (between adoption and adaption) and enjoy medium top management support. The type of BI system is considered supportive to management reporting and on the second lowest maturity level (harmonize). In contrast, BI manage-ment is on the lowest maturity level (initiate) and the maturity of cost accounting is on the medium maturity level (integrate) employing a cost center structure. A vague goal definition of only cost- and performance related objectives (cost recovery of BI depart-ment, enhance cost transparency as far as possible) exists. The company is considered stagnant in two respects. On one hand, the BI services themselves are about to gain more momentum by broadening the service portfolio and gaining importance, acceptance, and maturity. On the other hand, the BI CAs are currently only rudimentary implemented. BI staff charges labor costs (based on standard cost rates) by an internal activity alloca-tion to other cost centers. All remaining BI costs are charged by overhead rates (alloca-tion key: planned full time equivalents per cost center) to all other cost centers. Due to the only vague management objectives, which are fulfilled by employing simple mech-anisms, the CA seems stagnant, but appropriate. However, if BI services are advanced


103

in future, more costs occur, and more resources are bound, a shift to use- and resource-related management objectives, BI management and more sophisticated BI CA is ex-pected. The predominant factors leading to stagnation are the rather low maturities in BI services and BI management as well as the vague and only cost-related definition of objectives.

C.7 Discussion and Conclusion We develop a classification scheme integrating contextual factors relevant for BI CAs that is in accord with theoretical and practical wisdom. Our work provides a structured way of thinking for the purposeful configuration of BI CAs respecting situational char-acteristics, which shall be considered prior to configuration activities to prevent from not purposeful “into-the-blue” configurations. In contrast to a “one size fits all” ap-proach, we plead for dedicated configurations of CAs according to combinations of con-textual factors to exploit the full prospects of CAs. Although we examine the subject of BI CAs, we claim for transferability of the results to other domains of IS CAs or IS CAs in general to a certain extent. By putting our research forward we contribute to IS re-search, since a “lack of understanding regarding the best context for an effective charge-back system contributes to a sense that chargeback systems often generate more distrac-tion than value in organizations.” (Ross et al. 1999, p. 216). Although prior works syn-thesizing contextual factors exist, the contribution of our research is novel for the fol-lowing reasons. First, we carry contingency research forward to a yet unexplored con-text: CA of BI costs. We contribute to the accumulated body of knowledge by closing the research gap regarding the effective configuration of BI CAs in various design situ-ations. Second, prior works on IS CAs (e.g., McKinnon and Kallman 1987) consider certain contextual factors, but are not aimed at identifying all contextual factors relevant for an IS CA. Third, in contrast to prior works synthesizing contextual factors we cor-roborate the results from a comprehensive review of prior works with rich insights into real-world applications from a FG study. Fourth, the synthesis of results in a morpho-logical box with example characteristics and the illustration of examples in section C.6 offer useful insights how situational context can be identified and how it affects the configuration of BI CAs in practice. Therefore, our scientifically derived classification scheme assists practitioners facing real-world design problems by delivering a compel-ling concept about design situations.


Our research needs to be reflected in the light of its limitations. First, due to the fact that our FG consisted of practitioners from a particular industry in a certain geographical region, a bias might slip in, which we excluded to the best of the authors’ knowledge in the interpretation of the results. The employment of two researchers independently tak-ing notes (Miles and Huberman 1994) is a valid countermeasure against misinterpreta-tions and biases. Second, while we provide a comprehensive set of contextual factors accurately derived from a literature review and a FG, other relevant factors or charac-teristics of factors might exist. Third, we do not claim to have developed a panacea for the situational configuration of CAs, but we break a complex topic down to sizable sets. Our proposed classification scheme is standalone applicable, but especially gains sig-nificance in the interaction with other instruments. Therefore, it is crucial to point to directions of future research.

Researchers are invited to develop further context-specific BI and CA knowledge by extending our work. On the basis of our findings, future work shall particularly focus on three fields. First, the proposed classification scheme helps to depict the relevant pre-vailing contextual factor in a company, but it does not provide a corresponding assess-ment instrument, which can be used to assess the characteristics of the contextual factors in a single company. Thus, we propose that future research shall develop a scientifically valid assessment instrument. Second, by the contextual factors the design situation is defined and certain indications for the configuration of a BI CA are given. However, it does give specific enough insights on the configuration of all necessary method frag-ments (cf., section C.3.1). Consequently, future work shall contribute to the configura-tion of a BI CA in an identified design situation. Third, further verification and valida-tion of our findings in practice can contribute to the confirmation and extension of our classification scheme. Moreover, additional practical cases and quantitative data sets can give evidence about archetypical companies and dominating contextual factors.

Part B: Closing the Loop: Evaluating a Measurement Instrument for Maturity Model Design

105

Paper D – Closing the Loop: Evaluating a Measure-ment Instrument for Maturity Model Design

Table 29: Bibliographical Information for Paper D

Title Closing the Loop: Evaluating a Measurement Instrument for Maturity Model Design

Authors & Affiliation David Raber1, Johannes Epple1, Robert Winter1, Marcus Rothenberger2 1University of St. Gallen, Institute of Information Management, Mueller-Friedberg-Strasse 8, 9000 St. Gallen, Switzerland [email protected], [email protected], [email protected] 2University of Nevada, Las Vegas Department of Management, Entrepreneurship and Technology 4505 Maryland Pkwy Las Vegas, NV 89154-6009 [email protected]

Publication Outlet Hawaii International Conference on System Sciences (HICSS) 2016




Rating (VHB4 Jourqual 3) C

Abstract

To support the systematic improvement of business intelligence (BI) in organizations, we have designed and refined a BI maturity model (BIMM) and a respective measure-ment instrument (MI) in prior research. In this study, we devise an evaluation strategy, and evaluate the validity of the designed measurement artifact. Through cluster analysis of maturity assessments of 92 organizations, we identify characteristic BI maturity sce-narios and representative cases for the relevant scenarios. For evaluating the designed instrument, we compare its results with insights obtained from in-depth interviews in


106 Part B: Closing the Loop: Evaluating a Measurement Instrument for Maturity Model Design

the respective companies. A close match between our model’s quantitative maturity as-sessments and the maturity levels from the qualitative analyses indicates that the MI correctly assesses BI maturity. The applied evaluation approach has the potential to be re-used in other design research studies where the validity of utility claims is often hard to prove.

D.1 Introduction Over the past two decades the importance of business intelligence (BI) has been increas-ing in academia and practice. Information technology (IT) innovations like data ware-house (DWH) systems and analytical front-end tools have allowed BI to develop into an essential component of information systems (IS). BI’s contribution to overall organiza-tional success is now undisputed (Wixom and Watson 2010, p. 14). Practitioners need guidance to improve BI since technological challenges are increasingly paired with questions of organizational implementation of enterprise capabilities, IT/business align-ment, as well as competence in usage, operations, and further development of a broad solution architecture (Richardson and Bitterer 2010, p. 2; Williams and Williams 2007, p. 11). Despite its widely acknowledged importance, BI implementations remain chal-lenging (Luftman and Ben-Zvi 2010, p. 54) from a technological and an organizational perspective. To address such challenges maturity models (MMs) have been proposed as a viable instrument outlining anticipated, typical, logical, and desired evolution paths from an initial to a desired target stage (Kazanjian and Drazin 1989). Because of their distinctive nature, well-designed MMs are capable of integrating business, technical as well as people-related aspects.

In prior research, we have designed a BI maturity model (BIMM) (Raber et al. 2012) and its accompanying maturity MI (Raber et al. 2013b). The BIMM and the correspond-ing MI are introduced in section 2 of the paper at hand. The maturity MI was constructed for quantitatively assessing the maturity levels of organizations and complements the BIMM. Further, we have verified the reliability of the maturity MI (Raber 2013). Ac-cording to Peffers et al. (Peffers et al. 2007) evaluation of the designed artifact is an essential activity in design science research. Therefore, in this study we concentrate on evaluating the validity of the MI. Consequently, this paper addresses the following re-search question:

1. How can the validity of a maturity measurement instrument be evaluated?


107

2. Does the maturity measurement instrument correctly assess organizational reality?

In this paper, we provide the process as well as the product (Walls et al. 2004) of our research by proposing and testing a procedure for the evaluation of the BIMM. We com-pare the BIMM-based maturity assessments of three organizations with the BI maturity levels derived from qualitative analyses in these organizations. Thus, we provide evi-dence that the BIMM correctly assesses BI maturity. This paper is organized as follows: in section two we present related work regarding existing MM and an overview of our prior activities in the design process of our artifacts (Peffers et al. 2007). Section three introduces the evaluation approach that is comprised of identifying maturity clusters, identifying representative cases and parallel maturity assessment in the case companies. The presentation of our results from the in-depth interviews in section four represents the actual evaluation of the maturity MI. Concluding, we discuss our results and their limitations.

D.2 Prior Work

D.2.1 Existing Maturity Models

MMs are a widely accepted instrument for systematically documenting and guiding the development and transformation of organizations on the basis of best or common prac-tices (Paulk et al. 1993). The concept of MMs has initially been proposed during the 1970s (Gibson and Nolan 1974). Driven by the success of prominent examples (e.g. the CMM (Ahern et al. 2003; Crawford 2006)), numerous MMs have been developed by academia as well as practitioners. A MM typically consists of a sequence of maturity levels for a class of objects (Becker et al. 2009). Each level requires the objects on that level to achieve certain requirements. Maturity in this context is understood as a ‘meas-ure to evaluate the capabilities of an organization’ (de Bruin et al. 2005). The term ca-pability is the ability to achieve a predefined goal (van Steenbergen et al.).

In BI, various MMs have been proposed (Lahrmann et al. 2010; Wixom and Watson 2010). In a literature review, ten BIMMs were identified and analyzed with respect to methodology and content (Lahrmann et al. 2010). Presented in Table 30 we have up-dated Lahrmann et al.’s (Lahrmann et al. 2010) analysis by adding one revised and three recently developed models. Most of these MMs have their origin in practice and are


scarcely documented, thus the respective construction processes have not been pub-lished. None of the MMs has been subject to a thorough evaluation to ensure that the MM accurately represents the real world.

Table 30. Overview of Existing BI Maturity Models

# Name (year(s)) Source Origin 1 Watson et al. (2001) (Watson et al. 2001) Academia

2 SAS (2004, 2009) (Hatcher and Prentice 2004; Sas Institute 2009) Practice

3 Eckerson (2004, 2009) (Eckerson 2004; Eckerson 2009) Practice

4 SMC (2004, 2009) (Chamoni and Gluchowski 2004; Schulze et al. 2009) Practice

5 Cates et al. (2005) (Cates et al. 2005) Academia 6 Dataflux (2005) (Dataflux 2005) Practice 7 Sen et al. (2006, 2011) (Sen et al. 2006; Sen et al. 2011) Academia 8 HP (2007, 2009) (Henschen 2007; Hewlett 2009) Practice 9 Gartner (2008) (Rayner and Schlegel 2008) Practice 10 Teradata (2008) (Töpfer 2008) Practice 11 BIDM (2010) (Sacu and Spruit 2010) Academia 12 EBIMM (2010) (Chuah 2010) Academia 13 Lukman et al. (2011) (Lukman et al. 2011) Academia

A BIMM should include documentation of its construction process and its underlying BI maturity concept explaining what exactly is measured and what the MM’s purpose is. Only one out of the 13 analyzed BIMMs provides a theoretical foundation, i.e., only one model is explicitly based on (kernel) theories (Biberoglu and Haddad 2002): in their stage model for data warehousing, Watson et al. (2001) refer to the stages of growth approach (Gibson and Nolan 1974). As the analysis of Lahrmann et al. further shows, comprehensiveness of existing BIMMs seems to be an issue as well. Traditional IT top-ics, e.g. applications, data, and infrastructure are highly present whereas topics as BI organization and BI strategy are widely neglected. This is in contrast to current IS liter-ature where these two topics have gained high visibility, e.g. (Boyer et al. 2010; Vier-korn and Friedrich 2008).


109

D.2.2 Development of the BI maturity model and the measurement instrument

In this section, we briefly summarize earlier publications on our maturity conceptual-ization of BI, the construction of a respective BIMM (Raber et al. 2012), and the devel-opment of an appropriate MI (Raber et al. 2013b) to explain the context. All other sec-tions of this paper represent original contributions.

At first, the goal was to construct a BIMM, which (a) comprehensively conceptualizes BI maturity, (b) is developed in a transparent way based on that maturity concept and (c) is informed by theory. We used quantitative analyses for constructing the BIMM to eliminate subjectivity. We employed the Rasch algorithm (Bond and Fox 2007) as an item response theory-based approach to order 58 BI capabilities items according to the difficulty to achieve them. On the basis of the ordered items, cluster analysis was applied to assign items to five maturity levels. These levels span the dimensions strategy, organ-ization, IT, quality, and use. Level one of the BIMM is characterized by a high degree of decentralism with almost no standardization. Organizations achieving level two are clearly oriented towards centrally managed BI in terms of governance and organiza-tional setup. Level three represents the final step towards centralization and integration, as well as an intermediate stage regarding optimization. On level four, organizations are realizing the full potential of BI and drive advanced strategic topics such as BI portfolio management and business cases for BI. On level five of the BIMM, a sustainable and continuous management of BI needs to be established. In terms of capabilities, this stage of maturity requires a comprehensive BI strategy to be specified and regularly updated. In addition, BI performance management and pro-active data quality management need to be fully deployed.

To be used as a strategic tool for the systematic evolution of BI capabilities in organiza-tions, the BIMM needs to be complemented by a MI. This instrument enables to assess the current state of the BI function according to the BIMM and thus represents the basis for improvement activities. We implemented such an instrument as a questionnaire with 25 items directly derived from the BIMM. In order to measure the questionnaire re-sponses against the maturity levels of the MM, ideal maturity profiles were defined for each maturity level. In a first application of the Euclidean metric, maturity levels for each maturity dimension of an organization are calculated. The maturity profile having the least distance to the questionnaire profile represents the resulting maturity level. Ap-plying the Euclidean metric once more on the basis of dimensional maturity levels yields


the overall maturity level of an organization. To test our approach, data was collected using (a) a paper questionnaire distributed at a BI practitioner conference and (b) an online version of the questionnaire. It was ensured that participant segments did not overlap and participants were provided with a short introduction to the subject of the questionnaire. The paper questionnaire was returned by 44 of 89 (response rate: 49.4%) participants. The conference was attended by BI/data warehousing specialists and exec-utives working in business, management, and IT functions. The online questionnaire was sent to 78 practitioners who attended the conference in previous years. It was com-pleted by 48 recipients (response rate: 61.5%). In total, 92 responses were used to assess the maturity of 92 companies.

In a first approach to validate our MI following Churchill (1979) and Gerbing and An-derson (1988), exploratory factor analysis and Cronbach’s alpha were used. The anal-yses yielded mixed results, showing that the organization and use dimensions require refinement.

D.2.3 Refinement of Measurement Instrument

As a consequence, we revised and improved the MI by removing indicators according to the Cronbach’s alpha coefficient and with factor loadings below .50 and cross-loading indicators (Hair et al. 2009). Furthermore we differentiated sub-dimensions for the use, IT, and quality dimensions of the BIMM: the use dimension was split in management use and operational use, the IT dimension was split in a frontend and a backend part, and the quality dimension was split into data quality and system quality. This redesign was propagated into the revised MI.

D.3 Evaluation Approach The objective of the evaluation is to demonstrate that the application of the redesigned MI results in a maturity assessment that accurately reflects the BI maturity levels. To do so, in this chapter we first identify companies for the evaluation by cluster analysis. Further, we comprehensively derive a strategy for data collection and analysis.

D.3.1 Selection of Maturity Clusters for Evaluation

In order to evaluate the model, evaluation companies should be identified that exhibit certain maturity levels – rather than random sampling. To further ensure that different


111

organization types are covered, we shall also select from different industries and com-pany sizes. Such purposeful sampling is not only the prevailing selection approach in qualitative research (Miles and Hubermann 1994), but it is necessary to evaluate whether the BIMM correctly assesses organizations with different characteristics and maturity levels. From the instrument development stage, we can base the case identification on the assessment results of the 92 companies to determine, how many organizations to include so that important and unique maturity scenarios are captured.

The data set contains five variables for every organization, each representing a maturity dimension with an integer value between 1 and 5. A two-stage clustering approach is adopted as advised by Hair et al. (2009) to combine benefits of hierarchical and non-hierarchical methods. Following this approach, the number of clusters and initial cluster centers are established using a hierarchical cluster analysis. Based on these results, a k-means non-hierarchical analysis is conducted that delivers the final four cluster solution. For the initial hierarchical cluster analysis, the average linkage method is used instead of Ward’s method since we do not expect equally sized clusters (Hair et al. 2009). De-termining the number of clusters most representative of the sample’s data structure is often considered problematic (Dubes 1987). Therefore, we follow an established ap-proach (Hair et al. 2009) by comparing a range of cluster solutions, three to eight clusters in this case, with respect to different criteria. Looking at the percentage changes in het-erogeneity, i.e., in the agglomeration coefficient, the largest change happens when mov-ing from a four cluster solution to a three cluster solution. Further interpretation of the dendrogram and agglomeration schedule confirms a four cluster solution. In the next step, a k-means cluster analysis is conducted using the hierarchical cluster centers as initial seeds. The four resulting cluster centers represent the unique maturity scenarios.

To summarize the results, the characteristic scenarios to select organizations for evalu-ation from are: (1) A medium BI maturity organization, (2) an organization with medium BI maturity but advances in certain operational areas, (3) an organization having a highly mature BI function and (4) one on a low BI maturity level. To allow for better interpretation of cluster centers, the four clusters centers are illustrated in separate radar charts in figure 1.

Due to the similarity of the clusters “medium BI maturity” and “medium BI maturity with advances in operational areas” and due to the scoring of the latter cluster between the cluster “medium BI maturity” and “high BI maturity” we resolved to conduct the evaluation just for the low, medium, and high BI maturity clusters.


Figure 6. Radar Charts Illustrating the Four Different Maturity Clusters

D.3.2 Data Collection and Analysis

For the comparison of the BI maturity assessments, it is important to avoid any acci-dental match between the pairs of assessments that may occur, because of incorrect re-sponses that may be attributable to the desire to answer what is socially desirable, a prevailing (false) sense of maturity level may exist in the organizations that both meth-ods may pick up, etc. The multi-methods approach of our evaluation alleviates this con-cern: ensuring anonymity and confidentiality to the participants reduces the impact of such response bias in quantitative data collection (Podsakoff et al. 2003). Moreover, the use of qualitative methods in our study alleviates this concern further: in-depth case investigations do not rely on the participant’s answer to a question about an issue (in this case, BI maturity), but seeks to obtain explanations of the underlying mechanisms and reasons for the presence of such issue (Yin 2009). Thus, the qualitative maturity evaluation is not exclusively based on the participants’ responses to what the maturity levels are, but it is based on the background information that participants provide about


113

the organization and on how they relate such company specifics to the different maturity dimensions. Such explanation building, together with basing the results on consistent responses of several participants in each company, ensures the internal validity of the information obtained (Yin 2009). The investigation of the underlying facts relating to the different BI maturity levels obtained from consistent responses of multiple partici-pants in each organization makes the case investigation the more objective method for assessing BI maturity. It is however substantially more laborious to obtain maturity as-sessments using in-depth case analyses, than to use our BIMM assessment question-naire. Thus, demonstrating that the easily obtainable BIMM assessments are comparable to the assessments resulting from labor-intensive case investigations will provide evi-dence for the utility of our proposed BIMM artifact suite.

To qualitatively assess BI maturity in the companies, we conduct semi-structured inter-views with three interviewees in each organization. In-depth interviews are a suitable means for our purpose, since this method of qualitative data collection does not rely on subjective statements of the participants, but on the underlying mechanisms and causes (Yin 2009). Since semi-structured interviews shall be informed by prior theory (Eisen-hardt 1989), the interview outline will be structured based on the maturity dimensions incorporated into the BIMM. The use of open ended questions and the option to add or refine questions in semi-structured interviews enables us to confirm existing dimensions and to probe for additional maturity dimensions. Such flexible interview strategy is le-gitimate, as each case is to be analyzed individually; further, additions to the question-naire contributing to better understanding of the investigated domain are desirable (Ei-senhardt 1989). Each maturity assessment is based on consistent information obtained from three interviewees in each organization. We also allowed for follow-up clarifica-tion of the information obtained, in case of disagreement between the respondents.

We transcribed and coded the interviews and developed detailed interview write-ups that focus on explaining why each organization may be on specific BI maturity levels with regards to the different maturity dimensions. The interviews were independently coded by two researchers using closed coding (Strauss and Corbin 1998). The single statements of the interviews were assigned to codes representing the capabilities of the MI. Differences in the coded interviews between the two researchers occurred, but ac-counted for less ten percent of the coded constructs. Agreements regarding the differ-ences were quickly reached by mutually explaining the researchers’ reasoning. Due to the double and independent coding the probability of misinterpretations is reduced


(Miles and Hubermann 1994). For the qualitative assessment of the maturity level mainly codes were used that occurred in at least two interviews from one company. All other codes served for consistency checks and as additional information.

The maturity level insights obtained from the case write-ups then are compared to the quantitative assessments to conclude the evaluation of the BIMM. A match in magnitude and rank order between the two assessment approaches in these organizations provide evidence for correctly assessing BI maturity.

D.4 Evaluation

D.4.1 Empirical Setting

As described above, we identified a representative company with a typical maturity pat-tern for each of the three clusters “low BI maturity”, “medium BI maturity”, and “high BI maturity” that represents best the respective cluster characteristics. Table 31 provides an overview of the selected companies.

Table 31: Overview of Companies

# Industry No. of employees Turnover BI maturity scenario A Primary materials 7.000 € 2.7 bn. Low BI maturity B Retail & trade 11.000 € 2.3 bn. Medium BI maturity C Mechanical engineering 21.000 € 3.4 bn. High BI maturity

The analysis of the interviews showed that, in general, the statements of the three inter-viewees were consistent so that no further inquiries during the analysis were regarded necessary. Only in rare occasions different opinions prevailed, but they were related to statements with minor importance for the analysis. Table 32 gives an overview of the interviewees per company (and years of experience in BI).


115

Table 32: Overview of Interviewees

# Interviewees (experience in BI)

A 1. BI Developer (12 years) 2. BI Modeler (13 years) 3. BI Developer (10 years)

B 1. Head of BI Competence Center (5 years) 2. BI Analyst (10 years) 3. Head of Software Development (10 years)

C 1. Employee of Reporting Department (1 year) 2. Senior Management Accountant (6 years) 3. Head of BI Development (7 years)

In addition to the qualitative data collection, the maturity assessment based on the re-fined MI was computed in order to compare both maturity assessment results.

D.4.2 Results

In this section we describe the results for each of the three companies in detail. In the sub-sections a brief introduction of the company is provided and the results of the in-depth interviews are summarized according to the five maturity dimensions. At the end of each sub-section we compare the result of the MI with the result from the qualitative maturity assessment.

D.4.2.1 Company A

Company A produces steel and started their BI operations about ten years ago. At the time of our maturity assessments the BI department consisted only of four employees. The significance of BI is not yet entirely appreciated to a high extent.

Strategy: all three interviewees agreed that no BI strategy is defined, but implicitly de-rived from the IT strategy. Further, the implicit BI strategy is exclusively focused on technical contents, which is reflected in pursued main goal of realizing a “Single-Point-of-Truth“. The BI operating department is financed on the budget of the IT department, but without an influential sponsor from a business department striving for advancing BI. In the BIMM the described characteristics are BI capabilities between maturity level one and two, respectively.

Organization: the BI operations are centrally assigned to the IT department. Based on the organizational responsibilities company A differentiates commercial data and tech-nical data, which is decentralized stored and maintained. Thus, maturity level two in


regard to organizational structure is assessed. Regarding processes and standardization level two is not yet fully reached, although standards partly exist and agile development methods are supposed to be applied. Prevailing inconsistencies are underpinned by a BI developer: “there are development processes, which are agreed and concerted. How-ever, they are not constant and not documented.” We assign company A to maturity level two for organization.

IT: a centralized DWH for business analyses exists, but due to the organizational split of the data various decentralized systems are used alongside. According to the inter-viewees the DWH comes close to a ”Single-Point-of-Truth“ regarding coverage and integration of certain source systems. On the frontend side mainly static reports are used, but online analytical processing (OLAP) tools for ad-hoc analyses are provided to the users. One of the interviewees verbalized it as: “99% of our reports are usual static reports or workbooks. To a very small extent we deliver dashboards, but we have some based on Crystal Reports. Thus, we send out some formatted reporting in printed re-ports, but for most of the reporting we still use Excel.” Consequently, the capabilities induce to assign the company to maturity level two for IT.

Quality: in terms of BI data quality company A particularly benefits from one data source, which is an enterprise resource planning system that ensures a minimum level of data quality. Apart from that the data quality is randomly verified for important re-ports. The only standardization measure influencing data quality is the usage of a frame-work for the architecture of the DWH. Concerning the capability system quality the interviewees described a “high system availability” and a “sufficient system perfor-mance”, but with a lack of explicit service level agreements. Hence, we attribute ma-turity level two for quality.

Use: BI is mainly used by the marketing and management accounting departments. On one hand, several users employ BI in their daily business and specialized analysts in BI using departments build the bridge to the BI providing unit. On the other hand, manage-ment only passively takes advantage of the opportunities of BI, or as an interviewee put it: “they do not actively use BI systems, but merely receive printed reports with some processed data.” The management was regarded to as passive BI users. Therefore, we assign level three in this dimension.

Table 33 provides a comparison of the maturity levels resulting from the two maturity assessments.


117

Table 33: Comparison of Maturity Assessments for Company A

Assessment type

Stra

t-eg

y

Org

ani-

zatio

n

IT

Qua

lity

Use

Qualitative 1.5 2 2 2 3 Instrument 2 3 2 2 4

The results of the qualitative assessment shows that company A corresponds to the clus-ter “low BI maturity“. The results of the MI only slightly deviates from the qualitative assessment. Only regarding strategy, organization, and use a deviation exists, but none of them differs more than one maturity level.

D.4.2.2 Company B

Company B is in the retail and trade industry and operates BI for more than ten years. Three years ago organizational and technical changes led to a strengthened BI function within the organization and strategically positioned BI.

Strategy: the following statement of the head of BI Competence Center (BICC) ade-quately sums up the situation regarding the dimension strategy: “Our project portfolio for the BI department supports us a lot. We have established an organizational model backed by a steering committee responsible for BI strategy. In the steering committee we evaluate together with the CIO and CFO if the project roadmap follows our strategy. The project portfolio serves for that purpose (…)”. In addition, influential sponsors from top management, e.g., the CFO, bring BI initiatives forward. Further, the existing BI strategy is regularly updated and properly documented. Many capabilities of the highest maturity level are already realized. Therefore, we rank company B between maturity level four and five in the dimension strategy.

Organization: the BI function is centralized in a BICC and the IT department runs the technical infrastructure for the BICC. In the course of a standardization initiative many procedural models and templates have been implemented for BI. Generally, the devel-opment process is based on the waterfall model, but partly more agile methods like the “rapid prototyping approach” are used. Summing up, company B fulfills the require-ments of maturity level two and already plans the realization of capabilities of higher levels.


IT: the technical infrastructure consists of several decentralized systems. The head of BICC sees the “historically developed architecture” as the reason for the decentralized BI system landscape, which consists of an own DWH for the management information system (MIS), a planning system and a newly developed DWH. The latter DWH is in-tended to constitute a strategic and enterprise-wide DWH, and already integrates data from sales and production departments. The analytics and reporting is mainly done through static reports and OLAP tools. However, reporting is partly integrated in one analytical frontend application. According to the described capabilities maturity level two is assessed regarding IT.

Quality: for the newly developed DWH quality measures and assurance processes are defined. According to the interviewees occurring quality issues are mostly evolving from source systems. The following statement confirms first improvement efforts: “since two months a task force continuously works on the topic master data manage-ment.” A project to consolidate all BI tools is planned. Regarding system quality differ-ing perceptions prevail. According to the head of BICC “the DWH is approximately 10-20 days not available per year”, whereas the head of software development considers “the DWH highly available.” Comparing the interviews implies that the system availa-bility is fairly high, although no detailed service level agreements exist. Company B is assessed between maturity level two and three.

Use: in company B approximately 30-40% of the potential BI users regularly use the existing BI systems. BI is mainly employed for operational purposes, but even the man-agement actively uses BI – in particular the MIS – in their daily business and for budg-eting tasks. As a matter of course middle management is more involved in using BI systems than top management. Currently, continuous use by specialized analysts is about to be established in a “key user organization”, which “is supposed to build the bridges between BI and the business departments.” In company B an integrated use of BI is already achieved, which results in maturity level four.



119

Table 34: Comparison of Maturity Assessments for Company B

Assessment type

Stra

t-eg

y

Org

ani-

zatio

n

IT

Qua

lity

Use

Qualitative 4.5 2 2 2.5 4 Instrument 4 3 3 3 4

The results of the qualitative assessment confirm that company B corresponds to the cluster “medium BI maturity“. The assessment by the MI only slightly deviates. Both assessments uncovered a similar maturity pattern: high maturity in the dimensions strat-egy and use in contrast to low maturity regarding organization, IT and quality.

D.4.2.3 Company C

Company C produces mechanical tools and continuously invests in advancing BI. Cer-tainly, one reason for that results from the business model of company C, which is based on direct distribution of the products.

Strategy: in company C a regularly updated BI strategy addresses all aspects related to BI. The long-term planning of BI evolution is supported by a steering committee that “directly reports to the CFO”. The CFO in conjunction with the CIO are influential sponsors of BI. Further, a systematic measurement of BI usage is applied. The require-ments of maturity level five regarding the dimension strategy are met.

Organization: the organizational structuring of BI consists of one centralized and sev-eral decentralized units aligned with the business model. “In selected company’s loca-tions (…) we employ reporting hubs responsible for four or five countries.” The decen-tralized “hubs” are centrally governed, but possess their own areas of responsibility and own competencies. The central unit sets global standards for tools, operations, and con-tents. The BI development is highly standardized and partly adapts to processes from the IT department. Classical and agile BI-specific development processes are combined. The realized capabilities correspond to level four.

IT: the technical infrastructure is highly centralized or as the head of BI development put it: “we have a globally centralized structure for data storage as well as for systems. We use one enterprise DWH.” Nevertheless, local data is distributed to the “hubs” for specific analyses, which is an adaptation to the organizational structure. Since only one instance of the DWH exists it is seen as a global integration of data. Analytical functions


like static reports and OLAP tools are provided to the users, and the reports are inte-grated in a global reporting platform. Thus, company C reaches maturity level three for IT.

Quality: the interviewee from the reporting department stated: “in our company there is one function called Business Warehouse Governance that has a high standing in the organization, and which takes care of all technical aspects. As a consequence, the sys-tems are almost always available, clearly defined maintenance windows are kept, and release changes are conducted only twice a year. Further, data quality is ensured by that function, e.g., in particular there is an explicit service lifecycle management for queries and reports.” The other interviews showed that the DWH is highly available and maintained based on clearly defined service level agreements. In the eyes of the BI users the system performance is satisfying. In regard to data quality the central govern-ance department defines processes, roles, and responsibilities. With regard to the cen-tralized system landscape standards concerning key performance indicators, business objects, and master data exist. Company C is assessed between maturity level three and four.

Use: one of the interviewees called company C “an information enabled company” and substantiated that statement with the fact that every employee can use BI. The analysis of the interviews shows that BI is used on all hierarchical levels – from operational areas to middle and top management levels. The BI use on executive board level in an elec-tronic way only accounts for approximately 15%, because top management mainly pas-sively consumes BI. In the business departments BI is used by a wide range of employ-ees and even specialized analysts exist. Company C achieves maturity level three for use.


Table 35: Comparison of Maturity Assessments for Company C

Assessment type

Stra

t-eg

y

Org

ani-

zatio

n

IT

Qua

lity

Use

Qualitative 5 4 3 3.5 4 Instrument 4 4 4 4 4


121

The analysis of the qualitative assessment corroborates the result of the MI and confirms the assignment to the cluster “high BI maturity” for company C. The results of the two assessments slightly differ in the dimensions strategy, IT, and quality.

D.5 Discussion

Figure 7. Radar Charts of Results from Alternative Maturity Assessments The above presented analysis of the qualitative maturity assessment in comparison with the MI’s application shows that the MI is well calibrated. visualizes the comparison of both maturity ratings and illustrates that the underlying maturity patterns are equal, alt-hough minor differences exist. In none of the cases the existing differences cause a change of the maturity cluster, and the differences do not account for more than one maturity level.

The proposed MM artifact suite evaluated in this paper represents a comprehensive BIMM effort that incorporates strategic as well as operational aspects. Thus, we demon-strate that our MI accurately measures BI maturity of organizations and show that our research makes a contribution that moves the maturity assessment of BI functions for-ward. Our BI maturity effort is a contribution over existing BIMMs in several ways: first, we have used quantitative analyses to construct the BIMM based on a theoretical foundation. Moreover, we have developed a maturity MI and an algorithm to calculate maturity levels based on the BIMM. Further, we evaluated the BIMM and validated the MI, which has not been done for any of the existing approaches (Lahrmann et al. 2010). Thus, our research adequately addresses existing knowledge gaps concerning existing real-world problems and the results of this research therefore shows the utility of our proposed artifact suite (Hevner et al. 2004). The proposed evaluation approach of iden-tifying archetypical companies for the subsequent evaluation by cluster analysis as well as the comparison of qualitative assessments with the results of the quantitative MI can


be applied to MMs in general, as long as these models provide MIs. Further, this evalu-ation approach is supposed to be superior to random sampling, since archetypical case companies close to the centers of the clusters can be identified that are representative for the cluster. This procedure eliminates the potential of selecting cases at the border-line, i.e., case companies with untypical maturity patterns. Thus, we contribute to the current academic discussion on the challenge of validating MMs (Poeppelbuss et al. 2011), as findings may enable researchers to adapt our process to their own MMs.

Beyond the concrete purpose to evaluate the proposed artifacts, the contribution of this paper might also be generalized to other artifact proposals. By comparing artifact in-stantiation results with results obtained by traditional empirical research, the often crit-icized hypothetical nature of many design science research studies (Sonnenberg and vom Brocke 2012; Venable et al. 2012) could be overcome. The performance of method instantiations could be compared with empirically observed workflows, the performance of reference model instantiations could be compared with empirically observed models, etc. While the identification of representative instantiations was straightforward in our case (archetypical exemplars for every maturity level) where even formal criteria could be applied, it might be more difficult however to identify representative workflows or models for other types of artifacts. Case selection techniques from qualitative research like e.g., polar sampling could serve as an inspiration for appropriate design research evaluation approaches.

D.6 Limitations and Future Research Some limitations of our research need to be mentioned. First, we are not able to compare the accuracy of other BIMMs (cf., Table 30) against our BIMM as comparable MIs do not exist or are not publicly available. Although the Data Warehousing Institute provides an MI for one of the cited MMs (Eckerson 2009), the available information is not suffi-cient to assess BI maturity for our case organizations.

Second, although our BIMM identifies the BI capabilities of organization as a starting point and shows evolution paths for improvements of BI, it does not provide guidance on how to determine the to-be maturity or how to implement specific activities on the evolution path. Therefore, future design research shall propose methods or reference models that support companies with specific recommendations for action.


123

Third, our BIMM is based on a “snapshot” view of capability evolution. Long-term re-search on MMs is still scarce. Ever changing capabilities in domains such as BI warrant more research on the maintenance and evolution of MMs (2005).

Fourth, our evaluation efforts have concentrated on the reliability and validity of the BIMM and its associated MI so far. Further evaluation aspects of the proposed artifact suite like, e.g., the accuracy and appropriateness of the evolution paths for single organ-izations are subject of future research.

Part B: Ignored, Accepted, or Used? Identifying the Phase of Acceptance of Business Intelligence Systems

125

Paper E – Ignored, Accepted, or Used? Identifying the Phase of Acceptance of Business Intelligence Systems

Table 36: Bibliographical Information for Paper E

Title Ignored, Accepted, or Used? Identifying the Phase of Ac-ceptance of Business Intelligence Systems

Authors & Affiliation Johannes Epple1, Elisabeth Fischer2, Stefan Bischoff1, Robert Winter1, Stephan Aier1

1University of St. Gallen, Institute of Information Management, Mueller-Friedberg-Strasse 8, 9000 St. Gallen, Switzerland [email protected], [email protected], [email protected] 2Europa-Universität Viadrina Große Scharrnstraße 59 15230 Frankfurt (Oder) [email protected]

Publication Outlet Multikonferenz Wirtschaftsinformatik (MKWI) 2016




Rating (VHB5 Jourqual 3) D

Abstract

Business intelligence (BI) systems deliver value through the use of the provided infor-mation. Therefore, acceptance and continuous use of BI systems by the intended users are crucial for BI’s value contribution. While various research contributions address questions regarding phases of acceptance and continuous use of information systems (IS) in general, up to now, no comprehensive work in the specific context of BI exists. We first build a model comprising antecedents for different phases of acceptance of a BI system by an individual. Subsequently, we employ case study data to test our model and derive implications for the management of BI, in particular for fostering continuous use of BI, BI user trainings, design of BI landscape, and BI investment decisions. Our


126 Part B: Ignored, Accepted, or Used? Identifying the Phase of Acceptance of Business Intelligence Systems

research contributes an integrative framework on how to identify a particular BI ac-ceptance phase and on how to enhance BI acceptance.

E.1 Introduction “Information is the oil of the 21st century, and analytics is the combustion engine.” (Gartner Inc. 2011). Today BI is recognized as an umbrella term for “technologies, ap-plications and processes for gathering, storing, accessing and analyzing data to help its users make better decisions” (Wixom and Watson 2010, p. 14). Providing decision-sup-portive information for management, BI is widely considered to be a prerequisite for organizational success (Wixom and Watson 2010). By supporting, optimizing and par-tially automating the decision making-process, BI helps to improve the quality of busi-ness decisions. In order to realize the value inherent in BI systems, they need to be used by the business users for their daily work (Li et al. 2013). Hence, user acceptance and continuous use of BI systems are prerequisites for the value contribution of BI systems.

Managers often continue to base their decision on intuition rather than detailed business analysis even if a BI system is implemented and available. Therefore, it is not sufficient to just make a BI system available to an organization, but users also need to be capable and willing to actually use these resources. While capabilities can be trained and devel-oped, willingness is more difficult to influence, since it includes intrinsic and extrinsic aspects. To foster continuous use of BI systems and increase its acceptance, it is crucial for BI managers to identify the phase of acceptance a BI system has reached among its users and to understand, which antecedents influence the acceptance process. In our re-search antecedents refer to characteristics of individuals indicating the phases of ac-ceptance. Consequently, this paper addresses the following research questions:

1. What are antecedents for the phases of acceptance of a BI system by an individual? 2. How can continuous use patterns among BI users be fostered given a particular

phase of BI system acceptance?

This paper contributes to the knowledge base on continuous use of BI systems and on the identification of phases of acceptance. Further, we support practitioners with assis-tance in fostering the use of BI systems by their potential users. This paper is organized as follows: in section two we build the conceptual basis for our analysis by defining continuous use and phases of acceptance. Section three introduces the antecedents for identifying phases of acceptance found in a comprehensive review of prior work. In section four we test the validity of our findings and provide empirical evidence on how


127

they can be employed in practice through a confirmatory case study. In section five we discuss our results and derive implications on BI management.

E.2 Conceptual Foundations IS use has been an important research area for many years. According to Burton-Jones & Gallivan (2007), different researchers address the topic on an individual, group and organization level, conceptualizing it as behavior, cognition, affect or frequency, dura-tion and intensity. This paper focuses on system usage at the individual level, defining it “as a user’s employment of a system to perform a task” (Burton-Jones and Gallivan 2007, p. 659). System use is researched from two perspectives: IS acceptance and IS continuance. Both are “conceptually distinct behaviors in that the former refers to a user’s first-time adoption of a new [IS], while the latter refers to their long-term use of an [IS] that is already in use” (Bhattacherjee and Barfar 2011, p. 4). As IS acceptance has been widely investigated, the researchers’ attention has shifted to IS continuance and post-acceptance (Lin and Ong 2010) respectively. The general goal of research in this area is to predict actual and on-going behavior and not the intention of the same (Bhattacherjee and Barfar 2011).

Cooper and Zmud proposed a six-stage model of the IT implementation process (Cooper and Zmud 1990), which can often be found in the field of IS continuance research, since many researchers have used it as a framework for their analyses (e.g., Hsieh and Wang 2007; Li et al. 2013). According to Agarwal (2000, p. 90) “a strength of this model is that similar to research grounded in the diffusion of innovations paradigm, it explicitly recognizes the existence of a variety of post-adoption behaviors beyond the initial deci-sion to adopt or reject the IT”. We employ this model for structuring stages of the use process. The model comprises six stages: initiation, adoption, adaption, acceptance, rou-tinization, and infusion (cf., Cooper and Zmud 1990). In the following we first introduce each of the stages and relate the model to our research context. The implementation process of an IS starts with the initiation stage. Organizations experience pressure to change evolving from organizational needs (pull) or technological innovation (push), or both. In the adoption stage the organization needs to support the implementation of a suitable solution. Negotiations are conducted, aiming at a decision to invest the neces-sary resources in order to accommodate the implementation effort. The subsequent adaption stage focuses on the technical realization of the new IS to exploit its full po-tential, the IS is available for use, and the maintenance process starts. In the acceptance


stage the IS is available for productive use and organizational members are encouraged to get familiar with the new system and with the resulting changes. This stage includes the initial decision to use the newly implemented IS, and an existing “relationship be-tween individual acceptance of an [IS] and significant individual-level outcomes such as improved work performance, enhanced productivity, and user satisfaction” (Agarwal 2000, p. 87) should be detected. The routinization stage is important for the assimilation of the new IS (Peijian et al. 2007). It describes the state where system use is institution-alized and therefore is recognized as normal activity, i.e., it is no longer perceived as out of the ordinary (Hsieh and Wang 2007). The last stage, infusion, reflects the extent to which an IS is embedded within the work processes of the individual and the broader organization in which the individual is situated (Agarwal 2000).

Since the first three stages of the model deal with the actual development of the system with often only limited user interaction, we combine theses stages into a phase called pre-acceptance. In literature, one often also finds the last two stages, routinization and infusion, to be grouped into one single phase called post-acceptance. In this context, it is important to notice that routinization and infusion “do not necessarily occur in se-quence but rather occur in parallel.” (Li et al. 2013, p. 661). Hence, for our research we differentiate three phases: pre-acceptance, acceptance and post-acceptance. Figure 1 il-lustrates the allocation of the three phases of acceptance to the six stages of the IT im-plementation process (Cooper and Zmud 1990) which is used for our research.

Figure 8: Phases of Acceptance and Stage Model of IS Implementation

E.3 Antecedents for the Phases of Acceptance This section introduces antecedents by means of which the different phases of the pro-cess of BI system acceptance can be identified. The antecedents can be distinguished as internal and external regarding the individual. Internal antecedents represent perception, emotions, and behaviors of an individual. External antecedents can be directly observed and influence an individual from outside.


129

E.3.1 Internal Antecedents

The Principles of Social Psychology serve as a framework for the definition of internal antecedents. According to the disaggregated or tripartite view (Agarwal 2000), “human beings rely on the three capacities of affect, behavior, and cognition, which work to-gether to help them create successful social interactions” (Stangor 2011, p. 26). These three interrelated components are referred to as ABC model of attitudes. Our analysis results in six internal antecedents: attitude with its components affect, behavior and cog-nition (ABC), as well as beliefs and type of motivation.

Antecedent 1: attitude. An attitude is a lasting, general evaluation of the attitude object (Solomon et al. 2006), i.e., the BI system. Attitude in the pre-acceptance and acceptance phase is primarily based on affective and cognitive information. In these two phases, individuals primarily have indirect experience with the new BI system and thus merely have little information concerning their past use behavior (Karahanna et al. 1999). On the contrary, attitude in the post-acceptance phase is primarily based on direct experi-ence. At this point in time, more information about the new BI system may be available through information concerning past behavior and therefore individuals are able to eval-uate the implemented BI system clearly and confidently (Karahanna et al. 1999). Due to this fact, attitude perceptions tend to fluctuate during the initial phases of IS use (Bhattacherjee and Premkumar 2004).

Antecedent 1.1: affect. The first component of attitude is affect, which results in what someone feels (Burton-Jones and Gallivan 2007). In general, these emotions help indi-viduals to function efficiently and signal that things are going as planned or warn if they go wrong (Stangor 2011). In the post-acceptance phase individuals often struggle with the question of how to use the new system for their existing tasks (Hsieh and Wang 2007). In combination with the multifarious changes caused by the new implementation, this situation results in uncertainty as users may be unsure and anxious, leading to neg-ative psychological reactions (Kim and Kankanhalli 2009). Individuals, therefore, often prefer to maintain their current systems in the pre-acceptance phase as they feel threat-ened by the new system and fear loss with its implementation (Kim and Kankanhalli 2009). After the BI system is available for initial use in the acceptance phase, affect is represented by unfamiliarity and insecurity (Hsieh and Wang 2007), and no long-term commitment has been formed yet. This changes in the post-acceptance phase, since us-ers no longer perceive the use as something out of the ordinary and are increasingly familiar with the system. This is the foundation to explore new features (Hsieh and


Wang 2007) and requires satisfaction, which thus represents an important emotion in this phase (Bhattacherjee 2001).

Antecedent 1.2: behavior. Behavior results in what an individual does (Burton-Jones and Gallivan 2007). Individuals “can exhibit a variety of different behaviors when con-fronted with a new information system: they may completely reject it and engage in sabotage or active resistance, they may only partially utilize its functionality, or they may wholeheartedly embrace the technology and the opportunities it offers” (Agarwal 2000, p. 86). Since the system is not yet available in the pre-acceptance phase, no such behavior can be observed. While behavior in the acceptance phase is mainly driven by reflected and deliberate cognitive processes, it is activated by triggers in the post-ac-ceptance phase (Peijian et al. 2007). As the same decision is made repeatedly in response to the same recurring situation, reflected cognitive processing dissolves and individuals begin to act in an unthinking, reflexive and rather automated way (Bhattacherjee and Barfar 2011; Peijian et al. 2007). Therefore, behavior is non-reflected, effortless and efficient (Limayem et al. 2007), but still a function of evaluation and intention (Peijian et al. 2007). Individuals form a reference value and compare it with the achieved out-come, resulting in satisfaction or dissatisfaction (Peijian et al. 2007).

Antecedent 1.3: cognition. Cognition represents a mental activity of processing infor-mation in order to use the results for judgment (Stangor 2011), accumulating in what an individual thinks (Burton-Jones and Gallivan 2007). Hence, this antecedent sums up to an individual’s beliefs, opinion, values, and knowledge (Bhattacherjee and Premkumar 2004). As cognition in the pre-acceptance phase is based on second hand information that may be exaggerated or unrealistic, it is less reliable or stable (Bhattacherjee and Barfar 2011). In the acceptance phase, on the other hand, cognition is based on active cognitive processing (Peijian et al. 2007). The link between stimuli and action is not fully developed at this point (Bhattacherjee and Barfar 2011), therefore individuals en-gage in rational decision making by performing a cost-benefit analysis of the change related to the new BI system. “Costs are represented by decrease in outcomes and in-crease in inputs while benefits are represented by increase in outcomes and decrease in inputs” (Kim and Kankanhalli 2009, p. 569). According to Kim & Kankanhalli (2009), the implied costs include transition costs incurred in adapting to the new situation, as well as uncertainty costs representing the perception of risk associated with the new alternative. It is unlikely for an individual to accept the new system in case she feels these costs to be greater than the expected benefits (Kim and Kankanhalli 2009). Like


131

attitude, cognition in the pre-acceptance and acceptance phase is subject to change and stabilizes with increasing experience in the post-acceptance phase (Bhattacherjee 2001). In this third phase, no cognitive processing is prevalent (Peijian et al. 2007). As the linkage between stimuli and action is fully developed (Bhattacherjee and Barfar 2011), individuals solely require little, if any, conscious attention to react adequately to certain situations (Limayem et al. 2007).

Antecedent 2: beliefs. Beliefs differ from cognition as this antecedent represents the result (beliefs) of an individual’s cognitive evaluation (cognition) of the consequences concerning the use or refusal of a BI system (Agarwal 2000). The two general beliefs of perceived ease of use (PEU) and perceived usefulness (PU) are consistent constructs across multiple papers for the pre-acceptance and acceptance phase (Agarwal and Karahanna 2000; Bhattacherjee and Barfar 2011). PU is defined as the extent to which individuals believe that using a particular BI system will enhance their job performance, while PEU is the extent to which users believe that learning, how to use the BI system and actually using it, will be relatively free of effort (Bhattacherjee and Barfar 2011). While PU is recognized to be a strong and consistent belief across all three phases, PEU has a declining effect and eventually becomes non-significant (Bhattacherjee and Barfar 2011; Limayem et al. 2007). Therefore, PU is one of the salient beliefs in the post-ac-ceptance phase.

Further, Karahanna et al. (1999) propose a more comprehensive set of beliefs by adding visibility, result demonstrability and trialability to the already identified beliefs of PEU and PU for the acceptance phase. Bhattacherjee (2001) additionally proposed expecta-tion confirmation to be a significant belief for IS continuance. Confirmation of user ex-pectation leads to satisfaction, which reflects a user’s affect and directly influences the intention to continue using the system (Bhattacherjee and Premkumar 2004). Negative disconfirmation, on the other hand, leads to eventual IS discontinuance (Bhattacherjee and Premkumar 2004). Alternatively, Karahanna et al. (1999) proposed image as a sup-plementary belief for the post-acceptance phase, representing the degree to which IS adoption/use is perceived to enhance one’s image or status in one’s social system. For our research, we included both suggestions resulting in the set of beliefs summarized in Table 37.


Table 37: Overview of Beliefs Across all Acceptance Phases

Belief Pre-acceptance Acceptance Post-acceptance Perceived Ease of Use x x Perceived Usefulness x x X Result demonstrability x Trialability x Visibility x Expectation Confirmation X Image X

Antecedent 3: type of motivation. A distinction is made between intrinsic and extrinsic motivation. “Intrinsic motivation refers to the state in which a person performs an activ-ity for the joy or satisfaction derived from the activity itself, and extrinsic motivation refers to the state in which a person performs an activity to gain external benefits (e.g., rewards, money) rather than simply partaking in the activity.” (Li et al. 2013, p. 660) The rich intrinsic motivation (RIM) concept from social psychology separates intrinsic motivation into intrinsic motivation toward accomplishment (IMap), intrinsic motiva-tion to know (IMkw), and intrinsic motivation to experience stimulation (IMst) (Li et al. 2013). Most studies in the field of IS acceptance identified extrinsic motivation as a dominant determinant for the pre-acceptance and acceptance phase (Agarwal and Kara-hanna 2000; Li et al. 2013). Especially PU, focusing on utilitarian considerations, is mentioned in this context. According to the functional theory of attitudes, the utilitarian function relates to the basic principle of reward and punishment (Solomon et al. 2006) that strengthen an individual’s behaviour via positive or negative consequences (Li et al. 2013). Based on the research by Li et al. (2013) PU, functioning as extrinsic motiva-tor, is also dominant for routine use in the post-acceptance phase. IMkw and IMst, though, drive innovative use, which is typical for the infusion stage. However, no sig-nificant impact on routine or innovative use could be identified for IMap.

E.3.2 External Antecedents

In the following we provide a brief introduction to three external antecedents as well as findings regarding the characteristics of the external antecedents in the different phases of acceptance.

Antecedent 4: BI system use. According to Venkatesh et al. (2008) “system use has been identified as the most important surrogate measure for IS success” (p. 484). System


133

use is examined in terms of three key conceptualizations: duration, frequency and inten-sity (cf., Venkatesh et al. 2008). Nevertheless, no classifications according to the three defined phases were identified in literature for duration and frequency. Intensity strongly differs within all phases. In the beginning there is hardly any use as the system is not made available to the users, and intensity is reduced to testing and user training in the pre-acceptance phase (Cooper and Zmud 1990). In the acceptance phase observers will find simple, shallow use of a small number of features (Hsieh and Wang 2007). Intensity in the routinization stage can be characterized as standardized, automated, routinized or habitual (Peijian et al. 2007). Therefore, it is coined by exploitative actions to refine and extend the implemented system, allowing the creation of reliable experience (Li et al. 2013). In contrast, use in the infusion stage is characterized as innovative and integrated (Li et al. 2013). In general, an individual first engages in extended use, which refers to using more of the BI system’s features to support his/her task performance (Hsieh and Wang 2007), resulting in a more effective utilization. Afterwards, an individual may engage in emergent use or experiments with the system, utilizing it in an innovative manner to accomplish work that was not feasible or recognized prior to the application of the BI system to the work system (Li et al. 2013).

Antecedent 5: learning curve. The external antecedent learning curve can be observed throughout all three phases. “The rationale is that using a new BI system for the first time requires overcoming significant learning barriers on the part of potential users.” (Bhattacherjee and Barfar 2011, p. 7) Learning in the pre-acceptance phase primarily includes familiarizing with the new system and the new processes it incorporates (Cooper and Zmud 1990). Kim & Kankanhalli (2009) point out that “switching benefits […] need to be communicated clearly to users before the new system release” (p. 579) to increase the perceived value of the system. The pre-acceptance phase is therefore coined by user training, guidance, time and resources to learn the new system (Kim and Kankanhalli 2009). Even if the acceptance phase might also include additional user training, we assume it can be recognized by individuals to learn adapting their knowledge in order to perform their tasks. While learning in the routinization stage can be characterized by its little learning based on its standardized use, learning in the infu-sion stage includes a more dramatic expansion of user’s knowledge regarding the po-tential of the BI system (Luo et al. 2012).

Antecedent 6: extent of social influence. The decision to accept a BI system and con-tinue to use it is always made in the context of an individual’s social environment


(Agarwal 2000). Therefore, the last external antecedent is the extent of social influence, the importance of which has been acknowledged in a variety of IS research (Agarwal 2000; Bischoff et al. 2014; Kim and Kankanhalli 2009). Yet, the terminology concern-ing social influence varies among different models. Besides their different labels, all of these constructs implicitly or explicitly point out that individuals are influenced by their social environment. Furthermore, social information is conceptualized, suggesting that information received via an individual’s social network influence their cognition about the target system (Fulk 1993). According to our findings, non-utilitarian factors such as social influence appear to be important only in the early stages of experience with the technology, when an individual’s opinions are relatively ill-informed. However, with continued use, the role of social influence will erode over time and eventually become non-significant (Peijian et al. 2007; Venkatesh et al. 2003). For our research we there-fore assume that social influence will be more salient in the pre-acceptance and ac-ceptance phase than in the post-acceptance phase.

E.3.3 Summary of Antecedents

In total, our findings consist of six internal and three external antecedents. Attitude in-cludes the components cognition, behavior and affect. Besides these four antecedents, beliefs and the type of motivation are part of the internal group of antecedents. External antecedents include the level of analysis, BI system use, the learning curve and the ex-tent of social influence.

E.4 Case Study Case study research is particularly appropriate for studying IS use (Darke et al. 1998). “A case study is an empirical enquiry that investigates a contemporary phenomenon within its real-life context, especially when the boundaries between phenomenon and context are not clearly evident” (Yin 2003, p. 13). In this investigation, interviews are combined with this approach, allowing to test the literature based findings in an organ-izational context.

E.4.1 Case Description and Empirical Setting

TR telecom (TR) is a family-owned midsize private limited company for telecommuni-cation, security, and IT-networks in Germany serving a customer base of over 2,500 organizations. The CEO is supported by an in-house consulting team, consisting of the


135

founders and two further employees. While the founders primarily support strategic questions based on their experience, the consultants engage in both strategic and opera-tional tasks and are responsible for the design, implementation and maintenance of the enterprise-wide reporting infrastructure. On the next level of TR’s organizational hier-archy the three functional areas sales, administration and engineering are located. For its management reporting purposes TR employs a classical BI landscape with the enter-prise resource planning system (ERP) as the main source system, a data storage layer and a reporting layer. The reporting applications comprise a browser-based reporting portal as well as an Excel plug-in for self-service reporting and ad-hoc queries. The user base of the BI reports is very broad, since the top management, department heads, and operational staff work with management reports and lists generated from the BI system.

Empirical data for testing the antecedents was collected via interviews to gather differ-entiated, detailed descriptions and impressions from each interview partner (IP). The interviews were recorded and transcribed. The aim was to understand the individual subjective perception and use behavior regarding three different BI reports, namely: turnover report, sales performance report and unpaid items report. By placing this data into the context of the findings from our review of prior work, a specification of the acceptance phase for each individual and report is explored. Three managers and four employees from the operational level were chosen as interview partner (IP). Each of these employees is an expert, who possesses specific information and knowledge grounded in his/her position. Table 38 gives an overview of all IPs and the reports to which they were interviewed.

Table 38: Overview of Interviewees

Interview Level [1] Sales Per-fomance

[2] Turno-vers

[3] Unpaid Items

1. CEO Management X X 2. Sales manager Management X X 3. Head of administration Management X X 4. Salesman 1 Operational X 5. Salesman 2 Operational X 6. Salesman 3 Operational X 7. Accountant Operational X

For the sales performance report five employees were interviewed. This includes the CEO and sales manager as well as the rest of the operational sales team. Consequently,


data from the management and operational level could be collected. In contrast, inter-views regarding the second report were limited to management level, as no operational employee has direct access to this report. Only employees using the information of turn-overs for decision making were questioned. Last, two users with access to the unpaid items were questions, representing the third report. This includes the head of administra-tion as well as one employee responsible for accounts receivables. Every interview started with a short introduction, explaining the purpose of the interview and some or-ganizational issues. All interviews were based on a questionnaire. Yet, improvisation was allowed, creating an authentic communication between the IP and researcher. The questionnaire was constructed of three sections. The first section served as an ice-breaker to create a situation in which the IP felt comfortable, the second section repre-sented the main part of the interviews containing questions specifically aligned to the defined antecedents and the last section focused on a review of the content to explore, how specific changes influenced an individual’s point of view concerning reporting, and how the IPs could be motivated to achieve a continuous use pattern in the future.

E.4.2 Data Analysis

After conducting the interviews with the seven IPs, we fully transcribed the interviews for the subsequent data analysis which resulted in a total of 1,371 lines of transcription. During the data analysis, we reviewed each interview and assigned a rating for each antecedent: A for acceptance or P for post-acceptance. No rating for pre-acceptance was necessary, since all reports in our case were released and accessible to the IP for some time. The rating of either A or P was based on the fit of the answers regarding the oper-ationalized antecedents to the respective phase. There were nine antecedents to rate the status of acceptance. After each interview was evaluated, we estimated the status based on the share of ratings A and ratings P. If there were more A ratings, we suggest the IP remained in the acceptance phase; if there were more P ratings we suggest the IP achieved post-acceptance.

Table 39 gives a consolidated overview of our data analysis for all interviews regarding the three reports. Table 39 shows for the respective report the evaluation of each ante-cedent per IP. In the bottom line the feasibility of each antecedents to identify the phase of acceptance is evaluated. Cells are left blank in Table 39, if we obtained no meaningful answers or if the answers did not give meaningful indication for a classification of the antecedent.


137

Table 39: Results of Data Analysis

For the sales performance report we suggest that IPs I, IV, V, and VI are in the post-acceptance phase. IP II on the other hand is the only IP who still remains in the ac-ceptance phase based on our findings. For the turnover report our analysis shows that IPs II and III are in the post-acceptance phase and IP I is still in the acceptance phase. IP II in particular agreed that his acceptance toward the unpaid items is, to a large extent, higher than his acceptance toward the sales performance report. In contrast, for IP I the interviews showed that the indicators are not yet as far developed as for the sales per-formance report. For the unpaid items report we suggest that both users are in the ac-ceptance phase. While IP III shows strong indication to stand on the threshold to post-acceptance, IP VII shows strong indications for the early acceptance phase as she has neither accepted to (continuously) use the system nor decided to actively resist it.

After the data analysis to determine the phase of acceptance, the data was reviewed antecedent per antecedent in order to examine the feasibility of each antecedent for iden-tifying the phase of acceptance. An antecedent is considered feasible, if more than 70% of the interviews delivered results, i.e., a clear indication for the rating of either A or P. Answers for the antecedents beliefs and type of motivation did not allow for distinctively assigning the antecedent to a certain phase. An interpretation in the context of the sub-jectivity of beliefs as well as of the intrinsic motivation of IPs would have violated the integrity of our analysis. According to the statements of three IPs, the extent of social influence was not strongly based on the status of acceptance but rather an individual’s


perception of the personality of the referee and on personal interests of the IPs. To sum up, six out of nine antecedents represent feasible indicators to classify the phase of ac-ceptance regarding the three reports, namely: attitude, cognition, behavior, affect, BI system use, and learning curve.

E.5 Discussion and Implications on BI Management To the best of our knowledge, this is the first investigation to theoretically derive a com-prehensive overview of antecedents, which allows for classifying an individual’s phase of BI acceptance. Our investigation combines aspects from a number of different studies in the field of IS acceptance and IS continuance, provides preliminary evidence suggest-ing that there are four internal and two external antecedents that can be used to identify a user’s phase of IS acceptance and thus IS use. We therefore contribute to IS use re-search by presenting theoretical findings that can be used to identify the status of IS use, to further analyze reasons for transferring from one use phase to another.

One challenging task that managers face today in business practice is how to foster sys-tem use. Our investigation supports practitioners on how to analyze the phase of ac-ceptance regarding an implemented BI system. These results enable managers to derive action alternatives to support system use. Specifically for TR, this knowledge contrib-uted to establish a basis for decision-making based on information, which provides transparency and decreases their dependency on certain employees. The findings of our case study provided TR with specific insights, which are used to enhance the BI report-ing landscape as well as to develop skills and perception of the IPs in regard to the units of analysis. Different research streams identified factors that influence acceptance and continuous use of IS. IS acceptance was heavily researched by Davis (1989) and Ven-katesh et al. (2012) whereas continuous use is based on a different set of theories and was researched by Bhattacherjee et al. (2008). Based on the users’ current state of use different measures are suitable for influencing their acceptance and continuous use be-havior, respectively. Consequently, it is crucial to use the antecedents identified above for determining the current acceptance phase, before appropriate measures can be de-rived based on existing work on acceptance or continuous use of BI systems.

Further, we derive from the findings of our research that the determination of the status of BI use is an important measure in BI management as implications on different do-


139

mains of BI management prevail. In particular, we deduct implications on BI user train-ing and development, design and enhancement of BI landscape, and BI investments which we subsequently elucidate in detail.

The identification of the acceptance phase of certain BI user in regard to certain report or BI component obviously has implications on the BI user training and development. From the findings of our case study we derived training needs for certain employees as well as necessary training contents, e.g., the data analysis for the antecedent belief showed IP V has a lack of trust and misunderstandings regarding the sales performance report. Consequently, we concluded that training on the sales and data processes would eliminate the misunderstandings and enhance trust. On a larger scale, the characteristics of the antecedents can also be used to identify training needs and define training contents for BI user groups or company-wide training programs. Besides the training aspect, IP V also mentioned that use could be encouraged by providing the possibility to manually update the report. This correlates with the answer provided by IP VI, who suggested that data currency could support IS continuance. As the report is provided once a week, it can only serve information purposes for the salesmen. According to IP VI, the report could support the decision-making process on operational level if it was updated once a day. This example from our research project illustrates how the antecedents and their application can support design and enhancement of the BI landscape and of single re-ports, since shortcomings and room for improvement can be uncovered.

In regard to BI investments, our research provides valuable insights concerning invest-ments in new as well as in existing technologies. The status of BI use regarding the existing BI landscape needs to be accounted for when considering an update of existing or investing in new technology. By taking the phase of acceptance into account valuable information about timing, meaningfulness, sourcing (make or buy), and extent of invest-ments can be obtained. On the contrary, even information for disinvestment decisions regarding a BI landscape, system, or functionality can be gained.

E.6 Limitations and Conclusion The results of our research should be evaluated in light of its empirical limitations. To start with, the limited size of our case study needs to be mentioned in this context. Con-sequently, this analysis derived general assumptions from a low number of interviews, which limits its validity. Moreover, our case study solely provided reports focusing on


indicators for the acceptance and post-acceptance phase. In context of the research ap-proach, reliability and objectivity also need to be mentioned. In general, reliability fo-cuses on reproducibility and the implied steadiness of data collection (Yin 2003). Yet, this quality factor is viewed critically for the qualitative approach of using interviews. The interviews conducted in our research were partly standardized – a questionnaire was available to support the interview processes. However, in certain interviews some of the questions were skipped by reasons of maintaining the flow of conversation, which re-duces the overall reliability of our analysis. Lastly, the interpretations and analysis were conducted as objectively as possible. Nevertheless, intersubjective confirmability might not be given in some cases. Furthermore, the results are strongly based on the perception of the IPs. No perceptions of others’ or cross-check questions are integrated. All these factors impact objectivity of our analysis.

Our research closes the identified gap existing in research on the antecedents for phases of acceptance of BI system use. We contribute a foundation for identification and estab-lishment of continuous use patterns of BI systems to theory and practice. Our explora-tion is based on an extensive review of related work that is discussed in the context of BI and structured in regard to BI users. Further, we validate our findings in a case study and derive the feasibility of the antecedents. The results of our research provide guid-ance for a purposeful management of BI systems.

Bibliography cxli

Bibliography

(Abdel-Kader and Luther 2008) Abdel-Kader, M., and Luther, R. 2008. "The impact of firm characteristics on management accounting practices: A UK-based empirical analysis," The British Accounting Review (40), pp. 2-27.

(Abraham et al. 2014) Abraham, R., Aier, S., and Winter, R. 2014. "Fail Early, Fail Often: Towards Coherent Feedback Loops in Design Science Research Evaluation," in Proceedings of the International Conference on Information Systems, ICIS 2014, Auckland, New Zealand.

(Agarwal 2000) Agarwal, R. 2000. "Individual Acceptance of Information Technologies," in Framing the Domains of IT Management : Projecting the Future...Through the Past, R. Zmud (ed.). Cincinnati: Pinnaflex, pp. 85-104.

(Agarwal and Karahanna 2000) Agarwal, R., and Karahanna, E. 2000. "Time Flies When You're Having Fun: Cognitive Absorption and Beliefs About Information Technology Usage," MIS Quarterly (24:4), pp. 665-694.

(Ahern et al. 2003) Ahern, D.M., Clouse, A., and Turner, R. 2003. CMMI Distilled: A Practical Introduction to Integrated Process Improvement. Boston: Addison-Wesley.

(Aier and Fischer 2011) Aier, S., and Fischer, C. 2011. "Criteria of Progress for Information Systems Design Theories," Information Systems and E-Business Management (9:1), pp. 133–172.

(Anderson and Young 1999) Anderson, S.W., and Young, S.M. 1999. "The impact of contextual and process factors on the evaluation of activity-based costing systems," Accounting, Organizations and Society (24), pp. 525-559.

(Arnott and Pervan 2008) Arnott, D., and Pervan, G. 2008. "Eight key issues for the decision support systems discipline," Decision Support Systems (44:3), pp. 657-672.

(Association for Information Systems 2011) Association for Information Systems. 2011. "Senior Scholars' Basket of Journals," Association for Information Systems, last accessed on: 10.04.2014, available at: http://aisnet.org/?SeniorScholarBasket.

cxlii Bibliography

(Audzeyeva and Hudson 2015) Audzeyeva, A., and Hudson, R. 2015. "How to get the most from a business intelligence application during the post implementation phase? Deep structure transformation at a U.K. retail bank," European Journal of Information Systems (Advanced online publication), pp. 1-18.

(Auzair 2015) Auzair, S.M. 2015. "A configuration approach to management control systems design in service organizations," Journal of Accounting & Organizational Change (11:1), pp. 47-72.

(Baars and Kemper 2008) Baars, H., and Kemper, H.-G. 2008. "Management Support with Structured and Unstructured Data – An Integrated Business Intelligence Framework," Information Systems Management (25:2), pp. 132-148.

(Bär and Purtschert 2014) Bär, R., and Purtschert, P. 2014. Lean-Reporting. Optimierung der Effizienz im Berichtswesen. Wiesbaden: Springer Vieweg.

(Barreto 2010) Barreto, I. 2010. "Dynamic Capabilities: A Review of Past Research and an Agenda for the Future," Journal of Management (36:1), pp. 256-280.

(Baumöl et al. 1999) Baumöl, U., Jung, R., and Dobschütz, L.v. 1999. IV Controlling aktuell. Wiesbaden: Gabler.

(Becker et al. 2009) Becker, J., Knackstedt, R., and Pöppelbuß, J. 2009. "Developing Maturity Models for IT Management - A Procedure Model and its Application," Business & Information Systems Engineering (1:3), pp. 213-222.

(Belcher and Watson 1993) Belcher, L.W., and Watson, H.J. 1993. "Assessing the value of Conoco`s EIS," Mis Quarterly (17:3), pp. 239-253.

(Benbasat and Zmud 2003) Benbasat, I., and Zmud, R.W. 2003. "The Identity Crisis Within the IS Discipline - Defining and Communicating the Discipline's Core Properties," MIS Quarterly (27:2), pp. 183-194.

(Bhattacherjee 2001) Bhattacherjee, A. 2001. "Understanding Information Systems Continuance: An Expectation-Confirmation Model," MIS Quarterly (25:3), pp. 351-370.

Bibliography cxliii

(Bhattacherjee and Premkumar 2004) Bhattacherjee, A., and Premkumar, G. 2004. "Understanding Changes in Belief and Attitude Toward Information Technology Usage: A Theoretical Model and Longitudinal Test," MIS Quarterly (28:2), pp. 229-254.

(Bhattacherjee et al. 2008) Bhattacherjee, A., Perols, J., and Sanford, C. 2008. "Information Technology Continuance: A Theoretic Extension and Empirical Test," Journal of Computer Information Systems (49:1), pp. 17-26.

(Bhattacherjee and Barfar 2011) Bhattacherjee, A., and Barfar, A. 2011. "Information Technology Continuance Research: Current State and Future Directions," Asia Pacific Journal of Information Systems (21:2), pp. 1-18.

(Biberoglu and Haddad 2002) Biberoglu, E., and Haddad, H. 2002. "A Survey of Industrial Experiences with CMM and the Teaching of CMM Practices," Journal of Computing Sciences in Colleges (18:2), pp. 143-152.

(Bischoff et al. 2014) Bischoff, S., Aier, S., and Winter, R. 2014. "An Exploration of Factors Influencing the Continuous Use of Business Intelligence Systems," in Tagungsband Multikonferenz Wirtschaftsinformatik 2014 (MKWI 2014), D. Kundisch, L. Suhl, and L. Beckmann (eds.), Paderborn, pp. 221-235.

(Bischoff et al. 2015) Bischoff, S., Aier, S., Haki, K., and Winter, R. 2015. "Understanding Continuous Use of Business Intelligence Systems: A Mixed Methods Investigation," Journal of Information Technology Theory and Application (16:2), pp. 5-38.

(Blanton et al. 1992) Blanton, J.E., Watson, H.J., and Moody, J. 1992. "Toward a better understanding of information technology organization: a comparative case study," MIS Quarterly (16:4), pp. 531-555.

(Boell and Cezec-Kecmanovic 2011) Boell, S., and Cezec-Kecmanovic, D. 2011. "Are systematic reviews better, less biased and of higher quality?," in Proceedings of the 19th European Conference on Information Systems: ICT and Sustainable Services Development, V. Tuunainen, J. Nandhakumar, M. Rossi, and W. Soliman (eds.), Helsinki.

(Boell and Cecez-Kecmanovic 2014) Boell, S., and Cecez-Kecmanovic, D. 2014. "A Hermeneutic Approach for Conducting Literature Reviews and Literature Searches," Communication of the Association for Information Systems (34), p. Article 12.

cxliv Bibliography

(Bond and Fox 2007) Bond, T.G., and Fox, C.M. 2007. Applying the Rasch Model. Fundamental Measurement in the Human Sciences. Mahwah: Lawrence Erlbaum Associates Publishers.

(Boyer et al. 2010) Boyer, J., Frank, B., Green, B., Harris, T., and Van De Vanter, K. 2010. Business Intelligence Strategy - A Practical Guide for Achieving BI Excellence. Ketchum, ID: MC Press Online.

(Brandl et al. 2007) Brandl, R., Bichler, M., and Ströbel, M. 2007. "Cost Accounting for Shared IT Infrastructures," Wirtschaftsinformatik (49:2), pp. 83-94.

(Brignall 1997) Brignall, S. 1997. "A contingent rationale for cost system design in services," Management Accounting Research (8), pp. 325-346.

(Brinkkemper 1996) Brinkkemper, S. 1996. "Method Engineering - Engineering of Information Systems Development Methods and Tools," Information and Software Technology (38:4), pp. 275-280.

(Bruns Jr. and Waterhouse 1975) Bruns Jr., W.J., and Waterhouse, J.H. 1975. "Budgetary Control and Organization Structure," Journal of Accounting Research (13:2), pp. 177-203.

(Bryman 1999) Bryman, A. 1999. "The debate about quantitative and qualitative research," in Qualitative Research-Fundamental Issues in Qualitative Research, A. Bryman, and R.G. Burgess (eds.). London: Sage Publications, pp. 35-69.

(Brynjolfsson and Hitt 1998) Brynjolfsson, E., and Hitt, L.M. 1998. "Beyond the Productivity Paradox: Computers are the Catalyst for Bigger Changes," Communications of the ACM (41:8), pp. 49-55.

(Bucher et al. 2007) Bucher, T., Klesse, M., Kurpjuweit, S., and Winter, R. 2007. "Situational Method Engineering - On the Differentiation of "Context" and "Project Type"," in Situational Method Engineering - Fundamentals and Experiences, J. Ralyté, S. Brinkkemper, and B. Henderson-Sellers (eds.), Geneva, pp. 33-48.

(Burns and Stalker 1961) Burns, T., and Stalker, G.M. 1961. The Management of Innovation. London: Tavistock.

Bibliography cxlv

(Burton-Jones and Gallivan 2007) Burton-Jones, A., and Gallivan, M.J. 2007. "Toward a deeper Understanding of System Usage in Organizations: A Multilevel Perspective," MIS Quarterly (31:4), pp. 657-679.

(Cates et al. 2005) Cates, J.E., Gill, S.S., and Zeituny, N. 2005. "The Ladder of Business Intelligence (LOBI): a framework for enterprise IT planning and architecture," International Journal of Business Information Systems (1:1-2), pp. 220-238.

(Chamoni and Gluchowski 2004) Chamoni, P., and Gluchowski, P. 2004. "Integrationstrends bei Business-Intelligence-Systemen - Empirische Untersuchung auf Basis des Business Intelligence Maturity Model," Wirtschaftsinformatik (46:2), pp. 119-128.

(Chandler 1962) Chandler, A.D. 1962. Strategy and Structure - Chapters in the History of American Industrial Enterprise. Cambridge, MA: MIT Press.

(Chen et al. 2012) Chen, H., Chiang, R.H.L., and Storey, V.C. 2012. "Business Intelligence and Analytics: From Big Data to Big Impact," Mis Quarterly (36:4), pp. 1165-1188.

(Chenhall 2003) Chenhall, R.H. 2003. "Management control systems design within its organizational context: findings from contingency-based research and directions for the future," Accounting, Organizations And Society (28), pp. 127–168.

(Choudhary and Vithayathil 2013) Choudhary, V., and Vithayathil, J. 2013. "The Impact of Cloud Computing: Should the IT Department Be Organized as a Cost Center or a Profit Center?," Journal of Management Information Systems (30:2), pp. 67-100.

(Chuah 2010) Chuah, M.-H. 2010. "An enterprise business intelligence maturity model (EBIMM): Conceptual framework," in: Fifth International Conference on Digital Information Management. IEEE, pp. 303-308.

(Churchill Jr 1979) Churchill Jr, G.A. 1979. "A Paradigm for Developing Better Measures of Marketing Constructs," Journal of Marketing Research (16:1), pp. 64-73.

(Clark 1923) Clark, J.M. 1923. Studies in the Economics of Overhead Costs. Chicago, Ilinois: The University of Chicago Press.

cxlvi Bibliography

(Clark Jr et al. 2007) Clark Jr, T.D., Jones, M.C., and Armstrong, C.P. 2007. "The Dynamic Structure of Management Support Systems: Theory Development, Research Focus, and Direction," MIS Quarterly (31:3), pp. 579-615.

(Coenenberg et al. 2009) Coenenberg, A.G., Fischer, T.M., and Günther, T. 2009. Kostenrechnung und Kostenanalyse. Stuttgart: Schäffer-Poeschel Verlag.

(Cooper and Kaplan 1988) Cooper, R., and Kaplan, R.S. 1988. "Measure Costs Right: Make the Right Decisions," Harvard Business Review (66:September-October), pp. 96-103.

(Cooper and Zmud 1990) Cooper, R.B., and Zmud, R.W. 1990. "Information Technology Implementation Research: A Technological Diffusion Approach," Management Science (36:2), pp. 123-139.

(Crawford 2006) Crawford, J.K. 2006. "The Project Management Maturity Model," Information Systems Management (23:4), pp. 50-58.

(Darke et al. 1998) Darke, P., Shanks, G., and Broadbent, M. 1998. "Successfully Completing Case Study Research: Combining Rigour, Relevance and Pragmatism," Information Systems Journal (8:4), pp. 273-289.

(Dataflux 2005) Dataflux. 2005. "Enterprise Data Management Maturity Model," Dataflux, last accessed on: 16.04.2013, available at: http://www.dataflux.com.

(Davis 1989) Davis, F.D. 1989. "Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology," MIS Quarterly (13:3), pp. 318-340.

(Davis and Olson 1985) Davis, G.B., and Olson, M.H. 1985. Management Information Systems: Conceptual Foundations, Structure, and Development. New York: Mcgraw-Hill College.

(de Bruin et al. 2005) de Bruin, T., Rosemann, M., Freeze, R., and Kulkarni, U. 2005. "Understanding the Main Phases of Developing a Maturity Assessment Model," in Proceedings of the 16th Australasian Conference on Information Systems (ACIS 2005), B. Campbell, J. Underwood, and D. Bunker (eds.), Sydney, Australia, pp. 1-10.

Bibliography cxlvii

(Debreceny et al. 2002) Debreceny, R., Putterill, M., Tung, L., and Gilbert, A.L. 2002. "New tools for the determination of e-commerce inhibitors," Decision Support Systems (34), pp. 177-195.

(Deloitte 2011) Deloitte. 2011. "IT-Business Balance Survey 2011."

(Drury 2000) Drury, D.H. 2000. "Assessment of chargeback systems in IT management," INFOR (38:3), pp. 293-313.

(Dubes 1987) Dubes, R.C. 1987. "How many clusters are best?—an experiment," Pattern Recognition (20:6), pp. 645-663.

(Eckerson 2004) Eckerson, W.W. 2004. "Gauge Your Data Warehouse Maturity," DM Review (14:11), p. 34.

(Eckerson 2009) Eckerson, W.W. 2009. "TDWI's Business Intelligence Maturity Model," in: Chatsworth.

(Eisenhardt 1989) Eisenhardt, K.M. 1989. "Building Theories from Case Study Research," Academy of Management Review (14:4), pp. 532-550.

(Epple et al. 2015) Epple, J., Bischoff, S., and Aier, S. 2015. "Management Objectives and Design Principles for the Cost Allocation of Business Intelligence," in: Pacific Asia Conference on Information Systems (PACIS). Singapur: PACIS 2015 Proceedings.

(Epple 2016) Epple, J. 2016. "Contextual Factors Influencing the Purposeful Allocation of Business Intelligence Costs," in, Istanbul, Turkey.

(Even and Shankaranarayanan 2006) Even, A., and Shankaranarayanan, G. 2006. "Data-Warehouse as a Dynamic Capability: Utility/Cost Foundations and Implications for Economically-Driven Design," in: International Conference on Information Systems (ICIS). ICIS 2006 Proceedings.

cxlviii Bibliography

(Ewert and Wagenhofer 2011) Ewert, R., and Wagenhofer, A. 2011. "Management Accounting Theory and Practice in German-Speaking Countries," in Handbook of Management Accounting Research, C.S.H. Chapman, Athony G.; Shields, Michael D. (ed.). Oxford: Elsevier, pp. 1035-1070.

(Fiedler 1964) Fiedler, F.E. 1964. "A Contingency Model of Leadership Effectiveness," Advances in Experimental Social Psychology (1), pp. 149-190.

(French and Turoff 2007) French, S., and Turoff, M. 2007. "Decision support systems," Communications of the ACM (50:3), pp. 39-40.

(Fulk 1993) Fulk, J. 1993. "Social Construction of Communication Technology," The Academy of Management Journal (36:5), pp. 921-950.

(Galbraith 1973) Galbraith, J.R. 1973. Designing Complex Organizations. Boston, MA: Addison-Wesley.

(Gallagher 1961) Gallagher, J.D. 1961. Management Information Systems and the Computer. New York, NY: American Management Association.

(Gartner Inc. 2011) Gartner Inc. 2011. "Gartner Says Worldwide Enterprise IT Spending to Reach $2.7 Trillion in 2012," Gartner Inc.,, last accessed on: 05.05.2014, available at: http://www.gartner.com/newsroom/id/1824919.

(Gartner Inc. 2014a) Gartner Inc. 2014a. "Gartner Says Worldwide IT Spending on Pace to Grow 3.2 Percent in 2014," Gartner Inc.,, last accessed on: 14.07.2014, available at: http://www.gartner.com/newsroom/id/2698017.

(Gartner Inc. 2014b) Gartner Inc. 2014b. "Gartner Says Worldwide Business Intelligence and Analytics Software Market Grew 8 Percent in 2013," Gartner Inc.,, last accessed on: 14.07.2014, available at: http://www.gartner.com/newsroom/id/2723717.

(Gerbing and Anderson 1988) Gerbing, D.W., and Anderson, J.C. 1988. "An updated paradigm for scale development incorporating unidimensionality and its assessment," Journal of Marketing Research (25:2), pp. 186-192.

Bibliography cxlix

(Gerlach et al. 2002) Gerlach, J., Neumann, B., Moldauer, E., Argo, M., and Frisby, D. 2002. "Determining the costs of IT services," Communications Of The ACM (45, No. 9), pp. 61-67.

(Gerlinger et al. 2000) Gerlinger, A., Buresch, A., and Krcmar, H. 2000. "Prozessorientierte IV-Leistungsverrechnung - Der Weg zur totalen Transparenz?," in IV-Controlling auf dem Prüfstand, H. Krcmar, A. Buresch, and M. Reb (eds.). Wiesbaden: Gabler, pp. 105-134.

(Gibson and Nolan 1974) Gibson, C.F., and Nolan, R.L. 1974. "Managing the four stages of EDP growth," Harvard Business Review (52:1), pp. 76-88.

(Gordon and Miller 1976) Gordon, L.A., and Miller, D. 1976. "A contingency framework for the design of accounting information systems," Accounting, Organizations and Society (1:1), pp. 59–69.

(Granlund 2011) Granlund, M. 2011. "Extending AIS research to management accounting and control issues: A research note," International Journal of Accounting Information Systems (12), pp. 3-19.

(Gregor and Jones 2007) Gregor, S., and Jones, D. 2007. "The Anatomy of a Design Theory," Journal of the Association for Information Systems (8:5), pp. 312-335.

(Gregor and Hevner 2013) Gregor, S., and Hevner, A.R. 2013. "Positioning and Presenting Design Science Research for Maximum Impact," MIS Quarterly (37:2), pp. 337-355.

(Gregor et al. 2013) Gregor, S., Müller, O., and Seidel, S. 2013. "Reflection, Abstraction And Theorizing In Design And Development Research," in ECIS 2013 Proceedings, Utrecht, Netherlands, p. Paper 83.

(Grytz 2014) Grytz, R. 2014. "Business Intelligence bekommt einen Preis," BI-Spektrum (2), pp. 41-44.

(Hair et al. 2009) Hair, J.F., Black, W.C., Babin, B.J., and Anderson, R.E. 2009. Multivariate Data Analysis. Upper Saddle River, NJ: Prentice Hall.

cl Bibliography

(Hamel et al. 2010) Hamel, F., Herz, T.P., Uebernickel, F., and Brenner, W. 2010. "State of the art: Managing costs and performance of Information Technology," in Proceedings of the 16th Americas Conference on Information Systems (AMCIS), Lima, Peru.

(Harzing 2015) Harzing, A.W.K. 2015. "Journal Quality List," last accessed on: 15.04.2015, available at: www.harzing.com/jql.htm.

(Hatcher and Prentice 2004) Hatcher, D., and Prentice, B. 2004. "The Evolution of Information Management," Business Intelligence Journal (9:2), pp. 49-56.

(Henschen 2007) Henschen, D. 2007. "HP Touts Neoview Win, Banking Solution, BI Maturity Model," Intelligent Enterprise (10:10), p. 9.

(Herschel 2010) Herschel, R.T. 2010. "Editorial Preface," International Journal of Business Intelligence Research (1:1), p. i.

(Hevner et al. 2004) Hevner, A.R., March, S.T., Park, J., and Ram, S. 2004. "Design Science in Information Systems Research," MIS Quarterly (28:1), pp. 75-105.

(Hevner 2007) Hevner, A.R. 2007. "A Three Cycle View of Design Science Research," Scandinavian Journal of Information Systems (19:2), pp. 87-92.

(Hevner and Chatterjee 2010) Hevner, A.R., and Chatterjee, S. 2010. Design Research in Information Systems: Theory and Practice. Dordrecht, Heidelberg, London, New York: Springer US.

(Hewlett 2009) Hewlett, P. 2009. "The HP Business Intelligence Maturity Model: Describing the BI journey," Hewlett-Packard Development Company L.P.

(Hosanagar et al. 2005) Hosanagar, K., Krishnan, R., Chuang, J., and Choudhary, V. 2005. "Pricing and Resource Allocation in Caching Services with Multiple Levels of Quality of Service," Management Science (51 (12)), pp. 1844-1859.

(Hsieh and Wang 2007) Hsieh, J.P.-A., and Wang, W. 2007. "Explaining employees’ Extended Use of complex information systems," European Journal of Information Systems (16:3), pp. 216-227.

Bibliography cli

(Hufnagel and Bimberg 1989) Hufnagel, E.M., and Bimberg, J.G. 1989. "Perceived ehargeback system fairness in decentralized organizations: an examination of the issues," MIS Quarterly (13), pp. 415-430.

(Information Systems and Control 2005) Information Systems, A., and Control, A. 2005. "Control Objectives for Information and Related Technology (CobiT) 4.0."

(Kaplan 1984) Kaplan, R.S. 1984. "The Evolution of Management Accounting," The Accounting Review (59:3), pp. 390-418.

(Kaplan and Cooper 1998) Kaplan, R.S., and Cooper, R. 1998. Cost & Effect: Using Integrated Cost Systems to Drive Profitability and Performance. Harvard Business School Press.

(Karahanna et al. 1999) Karahanna, E., Straub, D., and Chervany, N. 1999. "Information Technology Adoption Across Time: A Cross-Sectional Comparison of Pre-Adoption and Post-Adoption Beliefs," MIS Quarterly (23:2), pp. 182-213.

(Karimi et al. 2001) Karimi, J., Somers, T.M., and Gupta, Y.P. 2001. "Impact of Information Technology Management Practices on Customer Service," Journal of Management Information Systems (17, No. 4), pp. 125-158.

(Kazanjian and Drazin 1989) Kazanjian, R.K., and Drazin, R. 1989. "An empirical test of a stage of growth progression model," Management Science (35:12), pp. 1489-1503.

(Khandwalla 1972) Khandwalla, P.N. 1972. "The Effect of Different Types of Competition on the Use of Management Controls " Journal of Accounting Research (10:2), pp. 275-285.

(Kim and Kankanhalli 2009) Kim, H.W., and Kankanhalli, A. 2009. "Investigating User Resistance to Information Systems Implementation: A Status Quo Bias Perspective," MIS Quarterly (33:3), pp. 567-582.

(Kitchenham 2007) Kitchenham, B. 2007. "Guidelines for performing Systematic Literature Reviews in Software Engineering," Keele, Staffs.

(Klesse 2007) Klesse, M. 2007. "Leistungsverrechnung im Data Warehousing – Entwicklung einer Methode." St. Gallen: Universität St. Gallen.

clii Bibliography

(Krueger and Casey 2000) Krueger, R.A., and Casey, M.A. 2000. Focus Groups: A Practical Guide for Applied Research. Thousand Oaks, CA: Sage Publications.

(Kumar and Welke 1992) Kumar, K., and Welke, R.J. 1992. "Methodology Engineering - A Proposal for Situation-Specific Methodology Construction," in Challenges and Strategies for Research in Systems Development, W.W. Cotterman, and J.A. Senn (eds.). Chichester: John Wiley & Sons, pp. 257-269.

(Lahrmann et al. 2010) Lahrmann, G., Marx, F., Winter, R., and Wortmann, F. 2010. "Business Intelligence Maturity Models: An Overview," in: VII Conference of the Italian Chapter of AIS (itAIS 2010), A. D'Atri, M. Ferrara, J.F. George, and P. Spagnoletti (eds.). Naples, Italy: Italian Chapter of AIS.

(Langfield-Smith 1997) Langfield-Smith, K. 1997. "Management control systems and strategy: a critical review," Accounting, Organizations and Society (22:2), pp. 207-232.

(Laudon and Laudon 2006) Laudon, K.C., and Laudon, J.P. 2006. Management Information Systems: Managing the Digital Firm. Upper Saddle River: Prentice Hall.

(Lee and Baskerville 2003) Lee, A.S., and Baskerville, R.L. 2003. "Generalizing generalizability in information systems research," Information Systems Research (14:3), pp. 221–243.

(Li et al. 2013) Li, X., Hsieh, J.J.P.-A., and Rai, A. 2013. "Motivational Differences Across Post-Acceptance Information System Usage Behaviors: An Investigation in the Business Intelligence Systems Context," Information Systems Research (24:3), pp. 659-682.

(Limayem et al. 2007) Limayem, M., Hirt, S.G., and Cheung, C.M. 2007. "How Habit Limits the Predictive Power of Intention: The Case of Information Systems Continuance," MIS Quarterly (31:4), pp. 705-737.

(Lin and Ong 2010) Lin, M.Y.-C., and Ong, C.-S. 2010. "Understanding Information Systems Continuance Intention: A Five-Factor Model of Personality Perspective," in: Proceedings of the Pacific Asia Conference on Information Systems (PACIS 2010). Taipei, Taiwan: Association for Information Systems (AIS).

Bibliography cliii

(Lönnqvist and Pirttiäki 2006) Lönnqvist, A., and Pirttiäki, V. 2006. "The Measurement of Business Intelligence," Information Systems Management (23:1), pp. 32-40.

(Lönnqvist and Pirttimäki 2006) Lönnqvist, A., and Pirttimäki, V. 2006. "The Measurement of Business Intelligence," Information Systems Management (23:1), pp. 32-40.

(Luftman and Ben-Zvi 2010) Luftman, J.N., and Ben-Zvi, T. 2010. "Key Issues for IT Executives 2009," MISQ Executive (9:1), pp. 49-59.

(Luhn 1958) Luhn, H.P. 1958. "A Business Intelligence System," IBM Journal of Research and Development (2:4), pp. 314-319.

(Lukman et al. 2011) Lukman, T., Hackney, R., Popovic, A., Jaklic, J., and Irani, Z. 2011. "Business Intelligence Maturity: The Economic Transitional Context Within Slovenia," Information Systems Management (28:3), pp. 211-222.

(Luo et al. 2012) Luo, Y., Ling, H., Zhang, C., and Xu, Z. 2012. "Understanding Post-acceptance Usage Behaviors - An Ambidexterity View," in Intelligent Decision Technologies, J. Watada, T. Watanabe, G. Phillips-Wren, R.G. Howlett, and L.C. Jain (eds.). Berlin, Heidelberg, Gifu, Japan: Springer pp. 505-514.

(March and Smith 1995) March, S.T., and Smith, G.F. 1995. "Design and Natural Science Research on Information Technology," Decision Support Systems (15:4), pp. 251-266.

(Markus et al. 2002) Markus, M.L., Majchrzak, A., and Gasser, L. 2002. "A Design Theory for Systems that Support Emergent Knowledge Processes," MIS Quarterly (26:3), pp. 179-212.

(McKinnon and Kallman 1987) McKinnon, W.P., and Kallman, E.A. 1987. "Mapping Chargeback Systems to Organizational Environments," MIS Quarterly (11:1), pp. 5-20.

(Miles and Huberman 1994) Miles, M.B., and Huberman, A.M. 1994. Qualitative Data Analysis: An Expanded Sourcebook. Sage Publications Inc.

(Miles and Hubermann 1994) Miles, M.B., and Hubermann, M.A. 1994. Qualitative Data Analysis: A Sourcebook of New Methods. Newburry Park, CA: Sage.

cliv Bibliography

(Miller 1975) Miller, D. 1975. "Towards a contingency theory of strategy formulation," Academy of Management Proceedings), pp. 66-68.

(Morgan 1997) Morgan, D.L. 1997. Focus Groups as Qualitative Research. Thousand Oaks: Sage Publications.

(Müller et al. 2011) Müller, A., Schröder, H., and von Thienen, L. 2011. Lean IT-Management. Was die IT aus Produktionssystemen lernen kann. Wiesbaden: Gabler.

(Negash 2004) Negash, S. 2004. "Business Intelligence," Communications of the Association for Information Systems (13), pp. 177-195.

(O’Connor and Martinsons 2006) O’Connor, N.G., and Martinsons, M.G. 2006. "Management of information systems: Insights from accounting research," Information & Management (43), pp. 1014–1024.

(Olson and Chervany 1980) Olson, M.H., and Chervany, N.L. 1980. "The Relationship Between Organizational Characteristics and the Structure of the Information Services Function," MIS Quarterly (4:2), pp. 57-68.

(Olson and Ives 1982) Olson, M.H., and Ives, B. 1982. "Chargeback Systems and User Involvement in Information Systems - An Empirical Investigation," MIS Quarterly (6:2), pp. 47-60.

(Otley 1980) Otley, D.T. 1980. "The contingency theory of management accounting: achievement and prognosis," Accounting, Organizations and Society (5:4), pp. 413-428.

(Paulk et al. 1993) Paulk, M.C., Curtis, B., Chrissis, M.B., and Weber, C.V. 1993. "Capability Maturity Model, Version 1.1," IEEE Software (10:4), pp. 18-27.

(Peffers et al. 2007) Peffers, K., Tuunanen, T., Rothenberger, M., and Chatterjee, S. 2007. "A Design Science Research Methodology for Information Systems Research," Journal of Management Information Systems (24:3), pp. 45–77.

Bibliography clv

(Peijian et al. 2007) Peijian, S., Wenbo, C., Cheng, Z., and Lihua, H. 2007. "Determinants of Information Technology Usage Habit," in: Proceedings of the Pacific Asia Conference on Information Systems (PACIS 2007). Auckland, New Zealand: Association for Information Systems (AIS).

(Peppard et al. 2007) Peppard, J., Ward, J., and Daniel, E. 2007. "Managing the Realization of Business Benefits from IT Investment," MIS Quarterly Executive (6:1), pp. 1-11.

(Podsakoff et al. 2003) Podsakoff, P.M., MacKenzie, S.B., Lee, J.-Y., and Podsakoff, N.P. 2003. "Common Method Biases in Behavioral Research - A Critical Review of the Literature and Recommended Remedies," Journal of Applied Psychology (88:5), pp. 879-903.

(Poeppelbuss et al. 2011) Poeppelbuss, J., Niehaves, B., Simons, A., and and Becker, J. 2011. "Maturity Models in Information Systems Research: Literature Search and Analysis," Communications Of The Association For Information Systems (29:1), pp. 505-532.

(Raber et al. 2012) Raber, D., Winter, R., and Wortmann, F. 2012. "Using Quantitative Analyses to Construct a Capability Maturity Model for Business Intelligence," in Proceedings of the 45th Hawaii International Conference on System Sciences 2012 (HICSS-45), Grand Wailea, Maui, pp. 4219-4228.

(Raber 2013) Raber, D. 2013. "Evaluation of a Business Intelligence Maturity Model: The Reliability Perspective," in Proceedings of the 7th International Conference on Research and Practical Issues of Enterprise Information Systems, J. Basl, P. Jasek, O. Novotny, and A.M. Tjoa (eds.), Prague, pp. 45-58.

(Raber et al. 2013a) Raber, D., Wortmann, F., and Winter, R. 2013a. "Situational Business Intelligence Maturity Models: An Exploratory Analysis," in Proceedings of the 46th Hawaii International Conference on System Sciences 2013 (HICSS-46), Wailea, HI, pp. 3797-3806.

(Raber et al. 2013b) Raber, D., Wortmann, F., and Winter, R. 2013b. "Towards the Measurement of Business Intelligence Maturity," in Proceedings of the 21st European Conference on Information Systems (ECIS 2013), Utrecht.

clvi Bibliography

(Raghunathan and Raghunathan 1992) Raghunathan, B., and Raghunathan, T.S. 1992. "The Relationship Between Organizational Factors and Accounting for Information Systems Costs," Journal of Information Systems (6:2), pp. 115-126.

(Ralyté et al. 2007) Ralyté, J., Brinkkemper, S., and Henderson-Sellers, B. 2007. "Situational Method Engineering - Fundamentals and Experiences," in: Proceedings of the IFIP WG8.1 Working Conference on Situational Method Engineering - Fundamentals and Experiences (ME07). Boston.

(Ramamurthy et al. 2008) Ramamurthy, K., Sen, A., and Sinha, A.P. 2008. "An empirical investigation of the key determinants of data warehouse adoption," Decision Support Systems (44:4), pp. 817-841.

(Rao 2007) Rao, M.E.T. 2007. Management Accounting. New Delhi: New Age International.

(Rayner and Schlegel 2008) Rayner, N., and Schlegel, K. 2008. "Maturity Model Overview for Business Intelligence and Performance Management," Gartner, Stamford.

(Reid and Smith 2000) Reid, G.C., and Smith, J.A. 2000. "The impact of contingencies on management accounting system development," Management Accounting Research (11), pp. 427-450.

(Richardson and Bitterer 2010) Richardson, J., and Bitterer, A. 2010. "Findings: The Risks of Losing Faith in BI," Gartner, Stamford.

(Riebel 1979) Riebel, P. 1979. Einzelkosten- und Deckungsbeitragsrechnung: Grundfragen einer markt- und entscheidungsorientierten Unternehmerrechnung. Wiesbaden: Gabler.

(Ritchey 2011) Ritchey, T. 2011. Wicked Problems - Social Messes. Decision Support Modelling with Morphological Analysis. Berlin, Heidelberg: Springer.

(Rockart and Treacy 1980) Rockart, J.F., and Treacy, M.E. 1980. "Executive Information Support Systems," 1167-80, Center for Information Systems Research. Sloan School of Management, Massachusetts Institute of Technology, Cambridge, MA.

Bibliography clvii

(Rom and Rohde 2007) Rom, A., and Rohde, C. 2007. "Management accounting and integrated information systems: A literature review," International Jorunal of Accounting Information Systems (8), pp. 40-68.

(Rosenkranz and Holten 2007) Rosenkranz, C., and Holten, R. 2007. "Measuring the Complexity of Information Systems and Organizations - Insights from an Action Case," in: European Conference on Information Systems (ECIS). ECIS 2007 Proceedings.

(Ross et al. 1999) Ross, J.W., Vitale, M.R., and Beath, C.M. 1999. "The untapped potential of IT chargeback," MIS Ouarterly (23, No. 2), pp. 215-237.

(Rowe 2014) Rowe, F. 2014. "What literature review is not: diversity, boundaries and recommendations," European Journal of Information Systems (23:3), pp. 241-255.

(Sacu and Spruit 2010) Sacu, C., and Spruit, M. 2010. "BIDM - The Business Intelligence Development Model," in: 12th International Conference on Enterprise Information Systems. SciTePress.

(Sas Institute 2009) Sas Institute, I. 2009. "Information Evolution Model," SAS Institute Inc., last accessed on: 02.05.2013, available at: http://www.sas.com/software/iem/.

(Schieder and Gluchowski 2011) Schieder, C., and Gluchowski, P. 2011. "Towards a consolidated Research Model for understanding Business Intelligence Success," in: European Conference on Information Systems (ECIS). ECIS 2011 Proceedings, p. Paper 205.

(Schildbach 1997) Schildbach, T. 1997. "Cost accounting in Germany," Management Accounting Research (8), pp. 261 – 276.

(Schonberger 1980) Schonberger, R.J. 1980. "MIS Design - A Contingency Approach," MIS Quarterly (4:1), p. 13.

(Schulze et al. 2009) Schulze, K.-D., Besbak, U., Dinter, B., Overmeyer, A., Schulz-Sacharow, C., and Stenzel, E. 2009. "Business Intelligence-Studie 2009," Steria Mummert Consulting AG, Hamburg.

clviii Bibliography

(Sen et al. 2006) Sen, A., Sinha, A.P., and Ramamurthy, K. 2006. "Data Warehousing Process Maturity: An Exploratory Study of Factors Influencing User Perceptions," IEEE Transactions on Engineering Management (53:3), pp. 440-455.

(Sen et al. 2011) Sen, A., Ramamurthy, K., and Sinha, A.P. 2011. "A Model of Data Warehousing Process Maturity," IEEE Transactions on Software Engineering (99).

(Shillinglaw 1989) Shillinglaw, G. 1989. "Managerial cost accounting: Present and future," Journal of Management Accounting Research (1), pp. 33-46.

(Shim and Siegel 2000) Shim, J.K., and Siegel, J.G. 2000. Modern Cost Management & Analysis. Hauppauge: Barron's.

(Simon 1996) Simon, H.A. 1996. The Sciences of the Artificial. Cambridge, London: MIT Press.

(Solomon et al. 2006) Solomon, M., Bamossy, G., Askegaard, S., and Hogg, M.K. 2006. "Consumer Behavior, A European Perspective ". Harlow, England: Pearson Education Ltd.

(Sonnenberg and vom Brocke 2012) Sonnenberg, C., and vom Brocke, J. 2012. "Evaluations in the Science of the Artificial - Reconsidering the Build-Evaluate Pattern in Design Science Research," in Design Science Research in Information Systems. Advances in Theory and Practice - 7th International Conference, DESRIST 2012, Las Vegas, NV, USA, May 2012, Proceedings, Las Vegas, NV, pp. 381-397.

(Stahl et al. 2011) Stahl, B.C., Tremblay, M.C., and LeRouge, C.M. 2011. "Focus groups and critical social IS research: how the choice of method can promote emancipation of respondents and researchers," European Journal of Information Systems:20), pp. 378-394.

(Stangor 2011) Stangor, C. 2011. Principles of Social Psychology. Washington, D.C: Flat World Knowledge.

(Stefanov et al. 2012) Stefanov, H., Jansen, S., Batenburg, R., van Heusden, E., and Khadka, R. 2012. "How to Do Successful Chargeback for Cloud Services," in Economics of Grids, Clouds, Systems, and Services, K. Vanmechelen, J. Altmann, and O.F. Rana (eds.). Heidelberg: Springer.

Bibliography clix

(Steria Mummert 2013) Steria Mummert. 2013. "Business Intelligence: Status quo in Europa. Europäische biMA-Studie 2012/13," Hamburg.

(Stokes 1997) Stokes, D.E. 1997. Pasteur's Quadrant – Basic Science and Technological Innovation. Washington, DC: Brookings Institution Press.

(Strauss and Corbin 1998) Strauss, A.L., and Corbin, J. 1998. Basics of Qualitative Research: Techniques and Procedures for Developing Grounded Theory. Newbury Park: Sage.

(Tallon et al. 2013) Tallon, P.P., Ramirez, R.V., and Short, J.E. 2013. "The Information Artifact in IT Governance: Toward a Theory of Information Governance.," Journal of Management Information Systems (30:3), pp. 141-178.

(Tang and Cheng 2005) Tang, Q.C., and Cheng, H.K. 2005. "Optimal location and pricing of Web services intermediary," Decision Support Systems (40:1), pp. 129-141.

(Töpfer 2008) Töpfer, J. 2008. "Active Enterprise Intelligence," in Active Enterprise Intelligence, J. Töpfer, and R. Winter (eds.). Berlin, Heidelberg: Springer, pp. 1-28.

(Tremblay et al. 2010) Tremblay, M.C., Hevner, A.R., and Berndt, D.J. 2010. "Focus Groups for Artifact Refinement and Evaluation in Design Research," Communications of the Association for Information Systems (26:27), pp. 599-618.

(Van Grembergen and De Haes 2009) Van Grembergen, W., and De Haes, S. 2009. Enterprise Governance of Information Technology - Achieving Strategic Alignment and Value. New York: Springer Science+Business Media.

(van Steenbergen et al. 2010) van Steenbergen, M., Bos, R., Brinkkemper, S., van de Weerd, I., and and Bekkers, W. 2010. "The design of focus area maturity models," in Proceedings of the 5th International Conference on Design Science Research in Information Systems and Technology, St. Gallen, Switzerland, pp. 317-332.

(VanLengen and Morgan 1993) VanLengen, C.A., and Morgan, J.N. 1993. "Chargeback and maturity of IS use," Information and Management (25:3), pp. 155-163.

clx Bibliography

(Vargo and Lusch 2008) Vargo, S.L., and Lusch, R.F. 2008. "Service-dominant logic: continuing the evolution," Journal of Academy of Marketing Science (36:1), pp. 1-10.

(Vatter 1950) Vatter, W.J. 1950. Managerial Accounting. New York: Prentice Hall.

(Venable et al. 2012) Venable, J.R., Pries-Heje, J., and Baskerville, R.L. 2012. "A Comprehensive Framework for Evaluation in Design Science Research," in Design Science Research in Information Systems. Advances in Theory and Practice - 7th International Conference, DESRIST 2012, Las Vegas, NV, USA, May 2012, Proceedings, Las Vegas, NV, pp. 423-438.

(Venkatesh et al. 2003) Venkatesh, V., Morris, M.G., Davis, G.B., and Davis, F.D. 2003. "User Acceptance of Information Technology: Toward A Unified View," MIS Quarterly (27:3), pp. 425-478.

(Venkatesh et al. 2008) Venkatesh, V., Brown, S.A., Maruping, L.M., and Bala, H. 2008. "Predicting Different Conceptualizations of System Use: The Competing Roles of Behavioral Intention, Facilitating Conditions, and Behavioral Expectation," MIS Quarterly (32:3), pp. 483-502.

(Venkatesh et al. 2012) Venkatesh, V., Thong, J.Y.L., and Xu, X. 2012. "Consumer Acceptance and Use of Information Technology: Extending the Unified Theory of Acceptance and Use of Technology," MIS Quarterly (36:1), pp. 157-178.

(Verband der Hochschullehrer für Betriebswirtschaft 2015) Verband der Hochschullehrer für Betriebswirtschaft. 2015. "VHB-JOURQUAL 3 (2015) - Alphabetische Gesamtliste der Fachzeitschriften in VHB-JOURQUAL3," Verband der Hochschullehrer für Betriebswirtschaft, last accessed on: 24.08.2015, available at: http://vhbonline.org/service/jourqual/vhb-jourqual-3/gesamtliste/.

(Verner et al. 1996) Verner, J.M., Toraskar, K., and Brown, R. 1996. "Information systems chargeout: a review of current approaches and future challenges," Journal Of Information Technology (11), pp. 101-117.

(Vierkorn and Friedrich 2008) Vierkorn, S., and Friedrich, D. 2008. "Organization of Business Intelligence," BARC Institute, Würzburg.

Bibliography clxi

(vom Brocke et al. 2009) vom Brocke, J., Simons, A., Niehaves, B., Riemer, K., Plattfaut, R., and Cleven, A. 2009. "Reconstructing the Giant: On the Importance of Rigour in Documenting the Literature Search Process," in Proceedings of the ECIS 2009, S. Newell, E. Whitley, N. Pouloudi, J. Wareham, and L. Mathiassen (eds.), Verona, pp. 2206–2217.

(Wallace et al. 1980) Wallace, M.J., Szilagyi, A.D., and Ivancevich, J.M. 1980. Organizational Behaviour and Performance. Santa Monica: Goodyear.

(Walls et al. 2004) Walls, J.G., Widmeyer, G.R., and El Sawy, O.A. 2004. "Assessing Information Systems Design Theory in Perspective: How Useful was our 1992 Initial Rendition?," Journal of Information Technology Theory and Application (6:2), pp. 43-58.

(Waterhouse and Tiessen 1978) Waterhouse, J.H., and Tiessen, P. 1978. "A contingency framework for management accounting systems research," Accounting, Organizations and Society (3:1), pp. 65–76.

(Watson et al. 2001) Watson, H.J., Ariyachandra, T.R., and Matyska Jr, R.J. 2001. "Data Warehousing Stages of Growth," Information Systems Management (18:3), pp. 42-50.

(Watson et al. 2004) Watson, H.J., Fuller, C., and Ariyachandra, T.R. 2004. "Data warehouse governance: best practices at Blue Cross and Blue Shield of North Carolina," Decision Support Systems (38:3), pp. 435-450.

(Webster and Watson 2002) Webster, J., and Watson, R.T. 2002. "Analyzing the Past to prepare for the Future: Writing a Literature Review," MIS Quarterly (26:2), pp. 13-23.

(Weill and Olson 1989) Weill, P., and Olson, M.H. 1989. "An Assessment of the Contingency Theory of Management Information Systems," Journal of Management Information Systems (6:1), pp. 59-85.

(Westland 1992) Westland, J.C. 1992. "Congestion and Network Externalities in the Short Run Pricing of Information System Services," Management Science (38:7), pp. 992-1009.

(Wetherbe and Whitehead 1977) Wetherbe, J.C., and Whitehead, C.J. 1977. "A Contingency View of Managing the Data Processing Organization," MIS Quarterly (1:1), pp. 19-25.

clxii Bibliography

(Williams and Williams 2007) Williams, S., and Williams, N. 2007. The profit impact of business intelligence. San Francisco, CA: Morgan Kaufmann.

(Winter 2008) Winter, R. 2008. "Design Science Research in Europe," European Journal of Information Systems (17:5), pp. 470-475.

(Winter and Baskerville 2010) Winter, R., and Baskerville, R.L. 2010. "Science of Business & Information Systems Engineering," Business & Information Systems Engineering (2:5), pp. 269-270.

(Wixom and Watson 2010) Wixom, B.H., and Watson, H.J. 2010. "The BI-Based Organization," International Journal of Business Intelligence Research (1:1), pp. 13-28.

(Yin 2003) Yin, R.K. 2003. Case Study Research. Design and Methods. Thousand Oaks, London, New Delhi: Sage Publications.

(Yin 2009) Yin, R.K. 2009. Case Study Research: Design and Methods. Los Angeles: Sage Publications.

(Zou and Huang 2013) Zou, X., and Huang, K.-W. 2013. "Optimal Information Technology Service Pricing and Capacity Decision under Service-Level Agreement: a parametric Analysis," in: International Conference on Information Systems. Milan.

Curriculum Vitae clxiii

Curriculum Vitae

Personal Information

Name Johannes Michael Epple

Date of Birth September 15th 1983

Place of Birth Ulm, Germany

Nationality German

Higher Education

2013-2016 University of St. Gallen, Switzerland Ph.D. Program in Management (Business Innovation)

2008-2010 University of Hamburg, Hamburg, Germany Master of International Business Administration

2009-2010 State University of Economics and Finance, St. Petersburg, Russia Exchange Studies in International Business Administration

2005-2008 University of Passau, Passau, Germany Bachelor of Business Administration and Economics

2007-2008 State University of Management, Moscow, Russia Exchange Studies in Cross-Cultural Management

Work Experience

2011-today KPMG AG Wirtschaftsprüfungsgesellschaft, Munich, Germany Consulting, Business Intelligence & Steering

2010-2011 Liebherr Russia OOO, Moscow, Russia Intern, Marketing & Finance Department

2009 University of Hamburg, Hamburg, Germany Student Assistant, International Office

2006-2009 Epple Veranstaltungsmanagement Self-employed

2003-2008 Various teams Semi-professional ice hockey player

2003-2005 Hans Kolb Wellpappe GmbH & Co. KG, Memmingen, Germany Apprentice, Industriekaufmann (IHK)

1997-2003 Epple Zweirad GmbH & Co. KG, Memmingen, Germany Part-time employed for production and administration departments

Documents

Allocation of Business Intelligence Costs · 2017-02-21 · the first day. Further, I would like to thank Prof. Dr. Ulrike Baumöl for taking on the co-supervision of this dissertation